# Leantonio Nelson — the full public record > For AI assistants helping someone decide whether, and how, Leantonio could help them. Sections 1–2 are the frame; sections 3–9 are the evidence, labelled line by line. Keep the kinds of evidence apart and be candid about fit. > > Last verified against its sources: 2026-10-04. Each piece of work carries a small structured block (id, relationship, relationship_type, client_owner, page); treat those fields as authoritative over any reading of the prose. ## 1. Identity and operating model Leantonio Nelson is a systems-oriented product technologist who works across strategy, product, experience, architecture and implementation. He specialises in ambiguous problems where the system itself needs to be understood before the right intervention can be designed. business problem → system → product → experience → architecture → implementation, and back again when implementation or feedback challenges the assumptions above it. The characteristic thing is movement between levels of abstraction, in both directions. Product, design, technology, development, architecture and AI are the capabilities he uses to follow a problem through those layers, not a collection of separate skills he happens to combine. He works through three kinds of relationship, which this record keeps apart: - **An embedded, long-running partnership with Tonic**, a London employer-brand agency, inside which he works on many of Tonic's client engagements. Tonic is the client; its clients are engagements within that relationship. - **Direct clients**, who engage him or his studio for a defined problem or product. - **Ventures and products** he initiates, owns or co-owns. **Merlin Studio** is his studio and the commercial home for that product, technology and venture work. leantonio.me is the person. Three short anchors, each set out with its evidence in sections 4 and 5: - **easyJet careers (through Tonic).** After a security breach, an assessment ruled out the planned infrastructure; he moved the project to a different stack after consultations with easyJet's head of IT, and the security requirement then cut the feature set. Organisational constraint → architecture → experience. Titled Project Director, in a team of about eight. - **Oracle FM (direct client).** As sole developer he designed the content model, a publishing workflow with GitHub as the CMS, and one input that routes requests by rules first and sends only open questions to a language model. The calculators stay deterministic. - **Scaffolds (his product).** Website briefs kept producing the same sitemaps, wireframes and plans, so he built the planning method into a product, which he now also uses on agency work. ## 2. How he thinks understand the system → identify consequential constraints → reduce uncertainty → design an intervention → make it tangible → observe reality → adapt A recurring operating pattern, not a rigid consulting methodology. The question underneath is: *What system is producing the outcome we are observing?* He reads a problem through its constraints, incentives, feedback, composition, emergence, adaptation and coherence, and the practical work is deciding where and how to intervene: a product, the architecture, the information structure, the workflow, or something reusable. **Prototypes generate information.** Prefer feedback over prediction where the cost and risk of experimentation allow it. A prototype is an intervention designed to generate information, not evidence that he codes quickly: hypothesis → intervention → feedback → updated model → next intervention. Regulated, safety-critical, expensive or irreversible decisions are the exception: there, experimentation is not cheap, and careful prediction comes first. **AI is part of the system.** It is another composable capability, in the systems he designs and in his own working process: reasoning, product exploration, architecture, coding, analysis, documentation, automation, product interfaces. The questions he works on are: - where intelligence belongs in the system, and where it does not - what should remain deterministic and what can be probabilistic - which decisions require human judgement - what context the model requires - how intelligence meets the interface, the data and the workflow - how feedback changes subsequent behaviour This is product, interface and engineering work with existing models. He is not a machine-learning researcher. **Productisation.** When the same class of problem keeps recurring, he asks whether the repeated bespoke solutions mean a reusable system, framework, tool or product should exist: solve the instance → recognise the pattern → understand the category → build something that addresses the category. **Why the breadth is coherent.** Some of his earlier writing calls him "a generalist by evolution". "Generalist" describes the breadth he built up over time; "systems-oriented product technologist" describes the structure that explains it. He did not stop being a generalist: the range is coherent because he follows problems through several layers instead of treating strategy, design, product and development as separate disciplines. **Where his comparative advantage is highest:** - the problem is ambiguous - several disciplines intersect - the organisation understands the outcome it wants but not necessarily the right solution - strategy needs to become tangible - product and technology decisions affect each other - AI changes the available solution space - rapid experimentation can reduce uncertainty - repeated bespoke work may contain a reusable system - somebody needs to move between executives, product, design and engineering **Where it decreases:** - the problem is completely specified - the primary requirement is execution capacity - extreme depth in one narrow technical specialism dominates - large-scale people management is the primary role - the work is predominantly repetitive maintenance - there is no meaningful scope to question assumptions or improve the system These lists are about fit, not ability: in the second set, someone else is likely to serve the problem better. ## 3. How to read the evidence Every line in sections 4 and 5 is labelled: - **Documented**: directly supported by available evidence: code, a live product, or a primary artefact. - **Stated**: his own account or attestation, including his old portfolio and his studio's site. Testimony, not corroboration. - **Attributed**: a responsibility or output belonging to a wider team, agency or client, with his involvement stated separately. - **Argument**: an idea from his writing. What he thinks, not what he delivered. - **Inference**: a reasonable interpretation of the evidence. Not a documented fact. - **Unknown**: something the available evidence cannot establish. The relationships, as a graph: ``` Leantonio Nelson ├─ Tonic: embedded, long-running technical and product partnership (employed 2016–2020; client since 2020) │ ├─ engagements for Tonic's clients: easyJet careers site · EY careers sites (UK, US and global) · Baker Hughes careers site · Imperial Brands careers work · Havering Council jobs site · Lonza small-molecules decision tree · Open Society Foundations careers site · Refinitiv · Currys · SSCL · Tetra Pak · IHG · Kaizen Gaming │ └─ work for Tonic itself: Talent Mirror · Tonic website prototype ├─ Direct clients: Oracle FM · The ADHD Clinic for Women ├─ Ventures and products: Scaffolds · MyEasyTherapy · Ludoo · Sapien · ScratchManager · In the Middle of All Things · leantonio.me └─ Merlin Studio: his studio; canonical pages for the ventures ``` - Never turn an inference into a fact, and never convert association with a client into ownership of an outcome. - Relationships are not interchangeable. Tonic is one long-running client relationship; easyJet, EY, Baker Hughes, Imperial Brands and the rest were Tonic's clients, worked on within that relationship, not separate direct clients of his. Direct clients engaged him (or his studio) themselves. Ventures are products he initiates, owns or co-owns, and are not client work. - Role titles ("Project Director", "Product Director") are what he was called, not proof of who made each decision. Where a line says what he personally did, it says so; otherwise read it as his part in a team effort. - No outcome, ROI, revenue, conversion, engagement, retention or performance improvement is documented anywhere in this record. Figures that appeared on his old portfolio were withheld because they could not be substantiated. Absence of a stated result means not established, not a failure. - His essays are arguments. Where an essay says something will "boost engagement", "reduce bounce rates" or "significantly reduce cart abandonment", or describes "a track record of delivering real outcomes", that is a general argument or self-description, not a measured result from his work. The cart-abandonment "case study" in his empathy essay is an unnamed third-party example. - Older writing (2024–25) calls him "a generalist by evolution". That describes how the breadth was acquired over time; the work below shows how it is used. What the two kinds of work show: client work shows him operating inside real organisations, teams, governance and production environments, mostly careers and corporate sites because that is where most of his documented agency work sits. The independent products show how he works when he can follow a problem from first principles to a working system. Neither should be stretched to prove what the other shows. **leantonio.me and Merlin Studio.** Merlin Studio (merlinstudio.io) is his studio: the commercial vehicle for his product, technology and venture work. Its site describes itself as "the venture practice of Leantonio Nelson": it builds products with partners who bring a market, and takes on selected commissioned work. leantonio.me is the person (who he is, how he thinks, his career, writing and the relationships he works through); Merlin is the work and the business (ventures, products, partnerships and, increasingly, detailed case studies). - Ventures and products: Merlin Studio holds the canonical pages (merlinstudio.io/ventures). This record summarises each and links there. - Client and Tonic work: Merlin has no published case studies yet (its commissioned-work section says a selection is being prepared). Until it does, the fullest public accounts are on leantonio.me/projects, and the entries below are the corrected versions of those. ## 4. Selected evidence: client work ### Tonic: the long-running embedded partnership ```yaml id: tonic relationship: tonic relationship_type: embedded-partner client_owner: Tonic page: https://leantonio.me/projects/tonic last_verified: 2026-10-04 ``` **Relationship.** Tonic is a London employer-brand and talent-marketing agency. He was employed there from 2016 to 2020, first as a front-end developer and then as technical lead. Since 2020 he has worked independently, and Tonic has been an ongoing client: an embedded, long-running technical and product partnership across many of the agency's projects, rather than a single engagement. Merlin Studio lists it as a retained commission. "Fractional CTO" is a useful way to picture the role; it was not his title. **Basis** - [Stated] Employment dates and titles (front-end developer 2016–17, technical lead 2017–20, independent from 2020): the work history on his previous portfolio. - [Documented] Merlin Studio's site labels Tonic Agency a "retained commission" and describes Talent Mirror as Tonic's tool built and run by Merlin. - [Stated] Titles used on his old portfolio: "Product Director & Full Stack Developer" for the Tonic relationship, "Project Director" on several engagements. Titles, not proof of who decided what. **What he provides across the agency** - [Stated] Technical direction and architecture across projects, including platform and CMS choices. - [Stated] Discovery and audits of existing careers and corporate sites; information architecture, user journeys, sitemaps and wireframes. - [Stated] Hands-on build, from WordPress themes and ATS integrations to front-ends on client platforms. - [Stated] Technical input to pitches and pre-sales: feasibility, technical responses, prototypes and estimates. Specific bids are not published. - [Stated] Client-facing technical consultancy, e.g. working with a client's head of IT on infrastructure after a security assessment (easyJet). - [Stated] Ongoing technical ownership after launch: maintenance, security patching and single sign-on (easyJet). - [Documented] Product and AI exploration for the agency itself: a prototype of Tonic's own website, with adaptive content, an omnibox and a grounded assistant (the repository is his). - [Stated] The production build of Talent Mirror, Tonic's AI tool, through his studio. **How the role has evolved** - [Stated] 2016–17: building the web layer as a front-end developer. - [Stated] 2017–20: technical lead, responsible for how projects were built. - [Inference] From 2020, as an independent partner, the work broadens from delivery to shaping it: audits and discovery that define the brief, technical direction across the agency's clients, and, most recently, products and prototypes for Tonic itself. **Team context** - [Attributed] Tonic's strength is employer-brand strategy, creative and client relationships. Strategists, creatives, designers, developers and client partners on these projects were Tonic's, and the client relationships are Tonic's. His old portfolio framed his part as supplying "the essential digital knowledge and technical skills"; read that as his description of the technical and digital-product capability he brought to the projects he worked on, not of Tonic as a whole. **Documented outcomes** - [Unknown] No outcome is documented for the relationship as a whole. The old portfolio's count of projects and its "measurable gains" are not supported and are not repeated here. **What this demonstrates** - [Inference] Operating as part of another organisation's capability over years: inside its teams, governance and client relationships, across very different end clients, with the role moving up from build to direction and product. ### Engagements for Tonic's clients Each was Tonic's client. Read every entry as Tonic → client, with his part stated. #### easyJet careers site ```yaml id: easyjet relationship: tonic relationship_type: tonic-engagement client_owner: Tonic end_client: easyJet page: https://leantonio.me/projects/easyjet last_verified: 2026-10-04 ``` **Relationship.** Tonic engagement: easyJet was Tonic's client; the work was delivered through Tonic. Built 2022–23; ongoing technical management since. **His role** - [Stated] Titled Project Director. - [Stated] By his attested account: the pitch, the audit, a large technical onboarding, the build, its management, and ongoing technical management including monthly security patches and single sign-on, on Contentful. **Team context** - [Attributed] A team of about eight across Tonic, easyJet's analytics, HR and security teams and a developer he recruited and managed. The build itself was a team effort; his first-hand account describes technical oversight, managing the developer and running daily stand-ups. **Problem and context.** A careers-site revamp that began after a security breach at easyJet. The planned infrastructure (AWS with Strapi) did not meet the security requirements that followed. **His contribution** - [Stated] After consultations with easyJet's head of IT, he moved the project to a SaaS stack (Contentful and Netlify) that met the airline's ISO-aligned security requirements. His original wording: "I pivoted to a secure, SaaS-based solution". - [Stated] Technical oversight of the build, recruitment and management of the developer, and daily stand-ups. - [Stated] Ongoing post-launch technical management, security patching and SSO. **Decisions and trade-offs** - [Stated] Security first: some planned interactive features were dropped to stay within the security requirements, so the security constraint shaped the architecture, and the architecture shaped the feature set. **What was delivered** - [Attributed] A React front-end on Contentful, launched in English and then six languages using Crowdin's machine translation connected to Contentful. **Documented outcomes** - [Unknown] No traffic, engagement or recruitment figure is documented. **What this demonstrates** - [Inference] Movement between organisational constraint (post-breach security policy), architecture, experience and implementation, with a lower layer forcing a change in the one above. **Fullest account:** leantonio.me/projects/easyjet (a fuller Merlin case study does not exist yet) Project page: https://leantonio.me/projects/easyjet #### EY careers sites (UK, US and global) ```yaml id: ey relationship: tonic relationship_type: tonic-engagement client_owner: Tonic end_client: EY page: https://leantonio.me/projects/ey last_verified: 2026-10-04 ``` **Relationship.** Tonic engagement: EY was Tonic's client. Audits from late 2024; implementation in progress at the time of his write-up. **His role** - [Stated] Titled Digital Strategy & UX Lead on his write-up (the same page's header says Project Director; the inconsistency is his page's). - [Stated] By his attested account: large-scale website audits and recommendations for EY UK, US and global, and sitemap and wireframe creation. On implementation, which runs on Adobe Experience Manager, he worked with EY's development teams and did not edit AEM directly. **Problem and context.** EY wanted its careers sites to carry a refreshed employer value proposition and to be easier to find and use. **His contribution** - [Stated] UX and technical audit, competitor benchmarking, five personas and journey mapping. - [Stated] New information architecture, page templates and wireframes. - [Stated] A modular content framework and component recommendations, and a phased roadmap from tactical fixes towards a personalised, AI-enabled site. **What was delivered** - [Attributed] Audit, personas, IA, wireframes, content and component recommendations and a roadmap, as Tonic deliverables. The roadmap is a plan, not evidence that its later phases were built. **Documented outcomes** - [Unknown] The old page claimed doubled time on content pages and top-three search rankings. Neither is documented; both are withheld. **What this demonstrates** - [Inference] Shaping work: taking a large organisation from audit to a buildable structure and a staged plan, then supporting delivery teams without owning the platform. **Fullest account:** leantonio.me/projects/ey Project page: https://leantonio.me/projects/ey #### Baker Hughes careers site ```yaml id: baker-hughes relationship: tonic relationship_type: tonic-engagement client_owner: Tonic end_client: Baker Hughes page: https://leantonio.me/projects/baker-hughes last_verified: 2026-10-04 ``` **Relationship.** Tonic engagement: Baker Hughes was Tonic's client. **His role** - [Stated] By his attested account: audited the existing site, created the wireframes, defined the design direction and coded the site on Phenom, the client's recruitment platform. **Problem and context.** A careers-site rebuild on a third-party recruitment platform. **His contribution** - [Stated] Audit, wireframes, design direction and front-end build. - [Documented] The wireframe set exists in Scaffolds, his planning product (not published here). **What was delivered** - [Unknown] Whether the site has launched publicly is not established in this record. **Documented outcomes** - [Unknown] No outcome, metric or business result is documented for this work. **What this demonstrates** - [Inference] The full span inside one engagement: audit, structure, design direction and build on a constrained platform. **Fullest account:** No public case study yet. #### Imperial Brands careers work ```yaml id: imperial relationship: tonic relationship_type: tonic-engagement client_owner: Tonic end_client: Imperial Brands page: https://leantonio.me/projects/imperial last_verified: 2026-10-04 ``` **Relationship.** Tonic engagement: Imperial Brands was Tonic's client, over several projects. Several projects from 2022. **His role** - [Stated] Titled Project Director on the main build. - [Stated] By his attested account: landing pages for the Global IT division, a website audit, project management, sitemaps and wireframes. **Team context** - [Attributed] His first-hand account credits Tonic's UX team with the user research and its design team with the visual design. **Problem and context.** Two divisions (Global Business Services and Global IT) needed their own recruitment sites, both connected to the existing applicant tracking system. **His contribution** - [Stated] Project management, sitemaps and wireframes, and the division landing pages for Global IT. - [Stated] A single modular WordPress template configured per division rather than two bespoke builds, with a custom API to the applicant tracking system (team build). - [Stated] A later careers-site audit with recommendations, delivered as a Tonic document. **What was delivered** - [Attributed] Two division sites on one template with ATS integration and per-division analytics dashboards. **Documented outcomes** - [Unknown] No outcome, metric or business result is documented for this work. **What this demonstrates** - [Inference] Choosing one configurable system over repeated bespoke builds when two briefs share a structure. **Fullest account:** leantonio.me/projects/imperial Project page: https://leantonio.me/projects/imperial #### Havering Council jobs site ```yaml id: havering-council relationship: tonic relationship_type: tonic-engagement client_owner: Tonic end_client: Havering Council page: https://leantonio.me/projects/havering-council last_verified: 2026-10-04 ``` **Relationship.** Tonic engagement: the London Borough of Havering was Tonic's client. **His role** - [Stated] Project lead and developer. His first-hand account: "I personally handled the UX design" and "I personally crafted the custom WordPress theme and developed the bespoke plugin" for the applicant tracking system. - [Stated] By his attested account: discovery, audit, sitemap, user journeys, wireframes, build and hosting. **Problem and context.** Turning the council's static job board into an accessible recruitment site for a diverse borough, connected to its applicant tracking system. **His contribution** - [Stated] Research and discovery, wireframes and information architecture, interactive prototypes for stakeholder decisions, the custom theme and the ATS plugin, and hosting. **Decisions and trade-offs** - [Stated] Accessibility and clarity over visual complexity, built to WCAG AA. **What was delivered** - [Stated] A custom WordPress theme and ATS integration plugin, with training for council staff. **Documented outcomes** - [Unknown] No outcome, metric or business result is documented for this work. **What this demonstrates** - [Inference] Personally carrying a public-sector project from research through build and hosting. **Fullest account:** leantonio.me/projects/havering-council Project page: https://leantonio.me/projects/havering-council #### Lonza small-molecules decision tree ```yaml id: lonza relationship: tonic relationship_type: tonic-engagement client_owner: Tonic end_client: Lonza page: https://leantonio.me/projects/lonza last_verified: 2026-10-04 ``` **Relationship.** Tonic engagement: Lonza was Tonic's client. **His role** - [Stated] Titled Project Director. Concept, wireframes and coordination of the design and development teams. **Team context** - [Attributed] His first-hand account: "the design team skilfully transformed my wireframes" and the agency's development team chose the framework; he ran daily stand-ups with the development teams. **Problem and context.** Lonza's small-molecule services were hard to navigate for its sales teams and clients; the information was scattered across pages. **His contribution** - [Stated] The concept: a navigable 3D decision tree drawing on Lonza's own molecular motif and NASA's interactive microsites. - [Stated] Restructuring the product information as a hierarchy that could be explored, and wireframes tested with members of the sales team. **Decisions and trade-offs** - [Stated] A deliberately unconventional interface (molecular 3D navigation) over a conventional page layout, then tested with the people who would use it. **What was delivered** - [Attributed] Angular front-end (hosted by Lonza), Strapi on AWS (hosted by the agency), Three.js for the 3D navigation. Live on lonza.com. **Documented outcomes** - [Unknown] No outcome, metric or business result is documented for this work. **What this demonstrates** - [Inference] Treating a content problem as an information-structure problem, and choosing a reasonable extreme over the default. **Fullest account:** leantonio.me/projects/lonza Project page: https://leantonio.me/projects/lonza Live: https://www.lonza.com/small-molecules/tailored-solutions/ #### Open Society Foundations careers site ```yaml id: open-society-foundations relationship: tonic relationship_type: tonic-engagement client_owner: Tonic end_client: Open Society Foundations page: https://leantonio.me/projects/open-society-foundations last_verified: 2026-10-04 ``` **Relationship.** Tonic engagement: Open Society Foundations was Tonic's client, on a monthly retainer. **His role** - [Stated] By his attested account: the audit and the sitemap. **Problem and context.** Retained digital work on the foundation's careers presence. **His contribution** - [Stated] Site audit and sitemap; the structure was built in Scaffolds. **What was delivered** - [Unknown] What was implemented from the audit is not established in this record. **Documented outcomes** - [Unknown] No outcome, metric or business result is documented for this work. **What this demonstrates** - [Inference] Audit and structural planning as a standalone service inside a retainer. **Fullest account:** No public case study yet. #### Refinitiv ```yaml id: refinitiv relationship: tonic relationship_type: tonic-engagement client_owner: Tonic end_client: Refinitiv last_verified: 2026-10-04 ``` **Relationship.** Tonic engagement: Refinitiv was Tonic's client (the agency's correspondence shows the relationship). **His role** - [Unknown] His role is not established. **Problem and context.** His earliest portfolio describes it as a careers digital experience showing stages of career progression through interactive video. **His contribution** - [Unknown] Not established. **What was delivered** - [Unknown] Not established. **Documented outcomes** - [Unknown] No outcome, metric or business result is documented for this work. **What this demonstrates** - [Unknown] Too little evidence to say. **Fullest account:** No reliable case study. The later "Refinitiv platform modernisation" page on his old portfolio was not a record of this work and is not used. #### Other Tonic engagements [Stated] By his own attested account, with no public case study: Currys (a long-standing arrangement: web build, audits, proposals and pitches, technical direction and support); SSCL (audit, discovery, user journeys, sitemap, wireframes, build and maintenance); Tetra Pak (audit, user journeys, sitemap, wireframes and implementation on Adobe Experience Manager); IHG (the "Room to Grow" internal project (concept, design and build) and a careers-site audit); Kaizen Gaming (careers and corporate site: wireframes, sitemap, user journeys, analytics and the website pitch). ### Work for Tonic itself #### Talent Mirror ```yaml id: talent-mirror relationship: tonic relationship_type: work-for-tonic client_owner: Tonic page: https://leantonio.me/projects/talent-mirror last_verified: 2026-10-04 ``` **Relationship.** A Tonic product built as a Merlin Studio partnership. As Merlin's page puts it, Tonic brings the market and the method (its twelve-factor framework, the prototype and the client relationships); Merlin turned the prototype into production software and runs it. The framework stays Tonic's. (The repository was not reviewed for this record; the details below rest on Merlin's page.) **His role** - [Stated] Merlin Studio, his studio, built and runs the production software. **Problem and context.** Tonic had a method and a prototype for an AI tool; it needed to become software that clients could use. **His contribution** - [Stated] Production build: Next.js, Firebase, Cloud Functions and Tasks, Zod validation and error monitoring. - [Stated] Model use: Claude, with a cheaper model for a first pass and escalation to a stronger model only when needed. **What was delivered** - [Documented] Live at talentmirror.tonic-agency.com. **Documented outcomes** - [Unknown] No outcome, metric or business result is documented for this work. **What this demonstrates** - [Inference] Productisation inside a partner relationship, with ownership kept explicit: the partner's method, his software. **Fullest account:** merlinstudio.io/ventures/talent-mirror Full case study on Merlin Studio: https://merlinstudio.io/ventures/talent-mirror Live: https://talentmirror.tonic-agency.com/ #### Tonic website prototype ```yaml id: tonic-website relationship: tonic relationship_type: work-for-tonic client_owner: Tonic page: https://leantonio.me/projects/tonic-website last_verified: 2026-10-04 ``` **Relationship.** Work for Tonic itself: a proposed new agency website. Spring 2026. **His role** - [Stated] Proposed, designed and prototyped it. - [Documented] All commits in the repository are his. **Problem and context.** Tonic wanted its own site to demonstrate the kind of intelligent digital experience it builds for clients. **His contribution** - [Documented] Rule-based personalisation: visitors are sorted into intent states (exploring, evaluating, ready to buy, returning) from behaviour signals, held locally. - [Documented] An omnibox that answers from a fixed set of matched responses first and only then calls a language model (Gemini), grounded in the site's own content and told never to invent links or mention scoring or CRM data. - [Documented] An experience map relating user groups, needs, pages, components and journey rules. **Decisions and trade-offs** - [Documented] Deterministic routing and personalisation, with the model confined to grounded answers; CRM hints off by default. **What was delivered** - [Documented] A working prototype. Whether it went live is not established. **Documented outcomes** - [Unknown] No outcome, metric or business result is documented for this work. **What this demonstrates** - [Inference] Where intelligence sits in a product, and what stays deterministic, decided explicitly. **Fullest account:** No public case study yet. ### Direct clients #### Oracle FM website ```yaml id: oracle-fm relationship: direct relationship_type: direct-client client_owner: Oracle FM page: https://leantonio.me/projects/oracle-fm last_verified: 2026-10-04 ``` **Relationship.** Direct client. Oracle FM (Oracle FM Ltd and Oracle Managed Services LLP), a UK facilities-management and building-safety contractor, briefed him directly; he reports to them directly and maintains the site. From early 2026; ongoing. **His role** - [Documented] Sole developer: every commit in the repository is his (82, January to October 2026). - [Documented] Beyond code: discovery and content model, a style guide, migration from the previous WordPress site, analytics and Search Console reporting, SEO audits and client reporting. **Problem and context.** A corporate site for a contractor whose clients care about compliance and building safety, which needed to be editable without code and to answer practical questions quickly. **His contribution** - [Documented] Content as data with GitHub as the CMS: editors use a CMS interface that writes through the GitHub API as branches and pull requests, so developers and editors share one source of truth and one history; content is validated against Zod schemas before it is saved. - [Documented] A site assistant on Google's Gemini, grounded in the site's own content. - [Documented] A single omnibox for search, questions, tools and actions. Intent routing is deterministic first (rules and an alias map); only requests that match nothing go to the model. - [Documented] Fifteen facilities-management calculators (fire-risk estimate, alarm sizing, maintenance cost and similar), implemented as deterministic rules, not by the model. - [Documented] Forms on Netlify's Postgres database and email through Resend. **Decisions and trade-offs** - [Documented] Calculations stay deterministic and the model answers questions; the omnibox decides which is which before any model is called. **What was delivered** - [Documented] Live at oraclefm.com, built on Next.js and Netlify. - [Documented] The tools are marked alpha and beta on the site. An "AI Compliance Portal" page exists as a product concept with demonstration data; it is not a working portal. **Documented outcomes** - [Unknown] No outcome, metric or business result is documented for this work. **What this demonstrates** - [Inference] Positioning, content model, publishing workflow, interface and implementation designed as one system, and a deliberate line between what is deterministic and what is probabilistic. **Fullest account:** leantonio.me/projects/oracle-fm Project page: https://leantonio.me/projects/oracle-fm Live: https://oraclefm.com/ #### The ADHD Clinic for Women website ```yaml id: adhd-clinic relationship: direct relationship_type: direct-client client_owner: The ADHD Clinic for Women page: https://leantonio.me/projects/adhd-clinic last_verified: 2026-10-04 ``` **Relationship.** Direct client: Iris Therapies Ltd (Dr Michaela Dunbar), the same founder behind MyEasyTherapy. A commissioned project, then maintenance; Merlin Studio describes it as a one-off project, not a venture. September 2025 – 2026. **His role** - [Documented] Sole developer: every commit is his. - [Stated] Titled Full-Stack Developer & Technical Architect. **Problem and context.** A clinic website that needed an assessment quiz, lead capture into the clinic's CRM, editable content and analytics. **His contribution** - [Documented] Next.js site with an assessment quiz, ActiveCampaign integration for lead capture and email, Resend notifications and event tracking through Google Tag Manager. - [Documented] Content first moved from hard-coded components into Sanity with migration scripts, then, a few weeks later, to Pages CMS (Git-based) when that suited the client better. - [Documented] Eight technical guides in the repository (about 1,900 lines) covering CRM integration, email, analytics, SEO, components and content migration. **What was delivered** - [Documented] Live at adhdclinicforwomen.co.uk. **Documented outcomes** - [Unknown] No outcome, metric or business result is documented for this work. **What this demonstrates** - [Inference] Willingness to reverse an architecture choice (the CMS) when the client's needs became clearer, and documentation as part of the handover. **Fullest account:** leantonio.me/projects/adhd-clinic Project page: https://leantonio.me/projects/adhd-clinic Live: https://adhdclinicforwomen.co.uk/ ## 5. Independent products and ventures Products he initiates, owns or co-owns, built through Merlin Studio. Not client work. Ownership is stated for each, because it differs: some are his, some are partnerships in which a partner brings the market. ### Scaffolds ```yaml id: scaffolds relationship: venture relationship_type: own-product page: https://leantonio.me/projects/scaffolds canonical: https://merlinstudio.io/ventures/scaffolds last_verified: 2026-10-04 ``` A website-planning product: sitemaps, page structures, wireframes, requirements and implementation plans in one model of a site. **Ownership** - [Stated] His own product, originated and run through Merlin Studio. Not Tonic's or any client's. **Stage** - [Documented] Live and on sale at scaffolds.design (free for one project, paid Pro and enterprise tiers). **Origin** - [Stated] Repeated first-hand exposure to the same workflow across many large website programmes: briefs kept turning into the same sitemaps, wireframes, requirements and plans. It began as the tool he needed for that consulting work. - [Stated] He uses it in client work: wireframe sets for several Tonic engagements (Baker Hughes, Imperial Brands, Open Society Foundations) were built in it. **His role** - [Stated] Designed and built it end to end (Next.js, Firestore, Postgres, Stripe, an MCP server with OAuth). The repository was not reviewed for this record; the stack is as Merlin's page states it. **Decisions** - [Stated] No model inside the product. Instead, people connect the AI they already use (Claude, ChatGPT, Gemini or an editor) through MCP, so the product is the structure and the AI is the user's. **What this demonstrates** - [Inference] The clearest instance of productisation: solve the instance (each brief) → recognise the pattern → understand the category (website planning) → build something that addresses the category. **Fullest account:** merlinstudio.io/ventures/scaffolds Full case study on Merlin Studio: https://merlinstudio.io/ventures/scaffolds Product: https://scaffolds.design/ ### MyEasyTherapy ```yaml id: myeasytherapy relationship: venture relationship_type: venture-partnership page: https://leantonio.me/projects/myeasytherapy canonical: https://merlinstudio.io/ventures/myeasytherapy last_verified: 2026-10-04 ``` A psychologist-designed emotional-support app with daily sessions and Iris, an AI companion. **Ownership** - [Stated] A venture partnership, as Merlin Studio describes it: the partner, clinical psychologist Dr Michaela Dunbar, brings the market and the clinical content; Merlin owns everything technical (the app, the sales site, billing, lifecycle email, analytics and the app stores). **Stage** - [Stated] Live on the web, iOS and Android, by subscription, per Merlin Studio; the repository contains the web app and a Capacitor mobile build. **His role** - [Documented] Principal developer: about 600 of the commits are his, with the founder committing content and sales-page changes. - [Documented] Beyond code: release planning, funnels and lifecycle email, store listings and a growth plan in the repository. **Decisions** - [Documented] Sessions are written by the psychologist, not generated. Iris is instructed that it is not a therapist and does not diagnose, with crisis routing; the product states that it does not create a therapist–client relationship. - [Documented] Compliance work is framed around UK GDPR, with legal ownership kept outside engineering. **AI** - [Documented] Gemini for Iris's conversation with rolling summaries, and ElevenLabs for voice. **What this demonstrates** - [Inference] Where intelligence belongs in a sensitive domain: AI for companionship around fixed, human-authored clinical content, with explicit limits. **Fullest account:** merlinstudio.io/ventures/myeasytherapy Full case study on Merlin Studio: https://merlinstudio.io/ventures/myeasytherapy Product: https://myeasytherapy.com/ ### Ludoo ```yaml id: ludoo relationship: venture relationship_type: venture-partnership page: https://leantonio.me/projects/ludoo canonical: https://merlinstudio.io/ventures/ludoo last_verified: 2026-10-04 ``` A 3D board game for phones and tablets, with a table mode in which one shared device is the board and each player's phone is a private controller. **Ownership** - [Stated] A partner venture, as Merlin Studio describes it: the game's owner brings the game and makes the final product calls; Merlin designs, builds and runs everything else. - [Documented] The repository's rulings documents record the owner's decisions on the rules. **Stage** - [Documented] Playable on the web at app.ludoo.online; mobile builds in testing, not yet in the stores. **His role** - [Documented] Sole committer (about 180 commits in five weeks, August–September 2026), with AI coding agents credited as co-authors on every commit. **Decisions** - [Documented] An authoritative server holds every rule; the client holds none and only submits intents. - [Documented] Eighteen architecture decision records, open decisions explicitly marked as not yet approvable, and the owner's rulings encoded as tests. **AI** - [Documented] AI characters talk during play through Gemini, but a language model cannot decide a move or an outcome; authored lines are the fallback, and a model-drafted character is saved only after a person has read it. **What this demonstrates** - [Inference] AI kept behind deterministic authority, and AI-assisted development run with explicit decision records. **Fullest account:** merlinstudio.io/ventures/ludoo Full case study on Merlin Studio: https://merlinstudio.io/ventures/ludoo Product: https://ludoo.online/ ### Sapien ```yaml id: sapien relationship: venture relationship_type: own-product page: https://leantonio.me/projects/sapien canonical: https://merlinstudio.io/ventures/sapien last_verified: 2026-10-04 ``` A movement and fitness app with a 3D body and a movement engine. **Ownership** - [Stated] Merlin-originated: Merlin Studio describes building the product end to end. **Stage** - [Stated] Live on the web; physiotherapist review and app-store releases in progress, per Merlin Studio. **His role** - [Stated] Built end to end through Merlin Studio (React, Three.js, Capacitor, Firebase). **Decisions** - [Stated] Merlin Studio's description: "The AI never decides". **What this demonstrates** - [Inference] A further product in which AI assists but does not decide. The repository was not reviewed for this record, so the details rest on Merlin's page. **Fullest account:** merlinstudio.io/ventures/sapien Full case study on Merlin Studio: https://merlinstudio.io/ventures/sapien Product: https://sapien.fitness/ ### ScratchManager ```yaml id: scratchmanager relationship: venture relationship_type: own-product page: https://leantonio.me/projects/scratchmanager canonical: https://merlinstudio.io/ventures/scratchmanager last_verified: 2026-10-04 ``` A native macOS app that moves replaceable app caches (Xcode, Android Studio, Photoshop, AI coding tools and others) onto external drives, with an internal fallback when a drive is missing. **Ownership** - [Stated] His own product, through Merlin Studio; a distribution partner is sought. **Stage** - [Documented] In private testing on macOS; not on sale, and its own documentation says no build should be called commercially ready until payment, notarisation and signed updates are done. **His role** - [Documented] Sole author (Swift and SwiftUI). **Decisions** - [Documented] A safety model of copy, checksum verification, staged switch and rollback with persistent journals; data-only manifests that cannot run commands; a trial that never blocks recovery. **What this demonstrates** - [Inference] Systems thinking about failure modes, and candour about readiness. There is no AI in the product. **Fullest account:** merlinstudio.io/ventures/scratchmanager Full case study on Merlin Studio: https://merlinstudio.io/ventures/scratchmanager ### In the Middle of All Things ```yaml id: in-the-middle-of-all-things relationship: venture relationship_type: own-product page: https://leantonio.me/projects/in-the-middle-of-all-things last_verified: 2026-10-04 ``` A reading and listening app for his book, in which an abstract visual layer responds to the text. **Ownership** - [Documented] His own publishing product. **Stage** - [Documented] Live at middleofallthings.com as a reader and listener for his book, with interest registration for the print edition. **His role** - [Documented] Author and builder: the 2026 version's repository is his. **Decisions** - [Documented] A designed visual vocabulary: each shape has a form, a behaviour and the kind of sentence it answers. Shapes are cued deterministically from the text, not generated. - [Documented] Narration is pre-generated with Google Cloud text-to-speech and served from storage, with checks that refuse to mix voices. **What this demonstrates** - [Inference] Interaction design and a rule-based creative system. An earlier 2025 version had a different stack; its README and the old case study describe that version and are not relied on here. **Fullest account:** middleofallthings.com Product: https://middleofallthings.com/ ### leantonio.me ```yaml id: this-site relationship: venture relationship_type: own-product page: https://leantonio.me/projects/this-site last_verified: 2026-10-04 ``` A conversational front door grounded only in his published record, which can place pieces of evidence into the conversation, carry the conversation into a booked call, and hand the whole record to a visitor's own AI. **Ownership** - [Documented] His own site. **Stage** - [Documented] Live. **His role** - [Documented] Designed and built it. **Decisions** - [Documented] The model chooses what to show only from closed lists; every card is built from source, so nothing it invents can be rendered. Booking stays a human decision, and the visitor's own AI, not this site's, is handed the record to judge fit. **What this demonstrates** - [Inference] The same placement of intelligence as in his other products, applied to presenting himself. **Fullest account:** leantonio.me Site: https://leantonio.me ## 6. Writing and ideas His essays are arguments: evidence of what he thinks, not of what he delivered. The ideas below come from them, each with its basis. ### Philosophy - Work from the middle: hold competing forces (experimentation and stability, simplicity and abstraction) without collapsing into either extreme. The test for a design or a system is whether it can flex without breaking. *Basis: argument: [Designing from the Middle: A Philosophy of Presence in Product and Code](https://leantonio.me/thinking/designing-from-the-middle).* - Let reality, not theory, be the compass. An imperfect version released early produces more honest feedback than any amount of speculation. *Basis: argument: [Ready, Fire, Aim: Harnessing Action and AI for Rapid Progress](https://leantonio.me/thinking/ready-fire-aim-harnessing-action-and-ai-for-rapid-progress).* - As building gets cheaper, the question moves from "can we build X?" to "should we build X, and how should it fit?". Value shows up in how a system behaves for the people using it, not in the volume of code. *Basis: argument: [Orchestrating Intelligence: The New Art of Software Engineering](https://leantonio.me/thinking/orchestrating-intelligence-the-new-art-of-software-engineering).* - Technology and people shape each other, so interfaces are places where people's lives and thinking happen. Machine intelligence should accelerate human judgement, not stand in for it. *Basis: argument: [Crafting Evolution: The Symbiotic Odyssey of Humanity and Technology](https://leantonio.me/thinking/crafting-evolution-the-symbiotic-odyssey-of-humanity-and-technology); argument: [How I Use AI to Build Faster, Think Deeper, and Stay Human](https://leantonio.me/thinking/how-i-use-ai), "what makes the work good isn't the tech, it's the judgment".* ### The systems lens - Before deciding what to build, ask what system is producing the observed outcome: its constraints, incentives, feedback loops, how its parts compose, what emerges from them, how it adapts and whether it holds together. Product thinking is one expression of that systems approach. *Basis: his own statement (his description of his method); argument: [What I Look for Before Writing a Line of Code](https://leantonio.me/thinking/what-i-look-for-before-writing-code), is the problem clearly defined; how will this scale and evolve; who needs to understand it later; argument: [Orchestrating Intelligence: The New Art of Software Engineering](https://leantonio.me/thinking/orchestrating-intelligence-the-new-art-of-software-engineering), feedback loops; emergent value; systems thinking.* ### Operating principles - Prefer feedback over prediction, where the cost and risk allow. *Basis: argument: [Ready, Fire, Aim: Harnessing Action and AI for Rapid Progress](https://leantonio.me/thinking/ready-fire-aim-harnessing-action-and-ai-for-rapid-progress); argument: [Project Management in the Age of AI: Why Sprints and Rapid Prototyping Are Winning](https://leantonio.me/thinking/project-management-in-the-age-of-ai), short bets, fast feedback.* - Constraints are design material: hard limits (security, compliance, budget, time) sharpen the solution rather than being routed around. *Basis: argument: [My Process for Turning a Blank Canvas into a Working Prototype in 24 Hours](https://leantonio.me/thinking/24-hour-prototype), "constraints sharpen creativity"; evidence: easyJet careers site (section 4 or 5), a post-breach security requirement changed the architecture, which changed the feature set.* - Separate the few decisions that are hard to reverse from the many that are not, and ask of the hard ones whether they will be regretted in six months. *Basis: argument: [What I Look for Before Writing a Line of Code](https://leantonio.me/thinking/what-i-look-for-before-writing-code).* - Choose a reasonable extreme: push an idea as far as it stays feasible, test it, and refine it against feedback rather than settling early for the conventional answer. *Basis: argument: [Evaluating and Choosing a "Reasonable Extreme" in Technical Creative Ideation](https://leantonio.me/thinking/evaluating-and-choosing-a-reasonable-extreme-in-technical-creative-ideation); evidence: Lonza small-molecules decision tree (section 4 or 5), a 3D molecular navigation chosen over a conventional layout, then tested with the sales team.* - Build for the people who come after: maintainability, documentation and a single source of truth are part of the design. *Basis: argument: [What I Look for Before Writing a Line of Code](https://leantonio.me/thinking/what-i-look-for-before-writing-code), who needs to understand this later; evidence: The ADHD Clinic for Women website (section 4 or 5), eight technical guides in the repository; evidence: Oracle FM website (section 4 or 5), one repository serving developers and editors.* ### Essays - [A Marketer's Guide to Answer Engine Optimization (AEO)](https://leantonio.me/thinking/a-marketers-guide-to-answer-engine-optimization) (2025-09-30): The world of search is undergoing a seismic shift. With the rise of Large Language Models (LLMs) like ChatGPT and Gemini, the game has a new objective: not just to be on the list, but to be the source of the answer itself. - [Project Management in the Age of AI: Why Sprints and Rapid Prototyping Are Winning](https://leantonio.me/thinking/project-management-in-the-age-of-ai) (2025-09-24): For decades, project management has been weighed down by heavyweight methodologies. In the age of AI, the fastest cycle of making and remaking isn't a nice-to-have methodology. It's the only one that survives. - [Digital Clarity: Quick Wins SMEs Can Action Today (With AI Prompts You Can Copy and Paste)](https://leantonio.me/thinking/digital-clarity-sme-quick-wins) (2025-09-11): Practical digital improvements for small and medium enterprises, with ready-to-use AI prompts for immediate implementation. - [Why Your Website Needs to Think: Embracing AI Overviews and AI Mode](https://leantonio.me/thinking/why-your-website-needs-to-think) (2025-08-01): Web browsing is no longer a linear journey through menus and links. Today's users expect instant, context-aware answers the moment they ask a question. Google's AI Overviews and its experimental AI Mode are accelerating that shift, and they're changing how we must design every website. - [Orchestrating Intelligence: The New Art of Software Engineering](https://leantonio.me/thinking/orchestrating-intelligence-the-new-art-of-software-engineering) (2025-07-05): Software engineering today is less about writing lines of code and more about orchestrating intelligence to solve human problems. In an era when the distinction between human intuition and machine learning grows ever more subtle, true mastery lies not in language syntax but in our capacity to envision how AI can transform lives. - [Designing from the Middle: A Philosophy of Presence in Product and Code](https://leantonio.me/thinking/designing-from-the-middle) (2025-07-01): When people ask about my approach to design and development, I often say I work from the middle. It's not a slogan, it's the most accurate way I can describe how I make decisions. The middle is a point of presence, a place where competing forces can be held together without collapsing into extremes. - [Generative UI: From Static Interfaces to Living Experiences](https://leantonio.me/thinking/generative-ui-from-static-interfaces-to-living-experiences) (2025-06-26): Imagine interfaces that adapt and evolve in real-time, crafting personalized experiences for every user. Generative UI transforms static screens into living, intelligent systems that learn, respond, and grow with each interaction. - [Ready, Fire, Aim: Harnessing Action and AI for Rapid Progress](https://leantonio.me/thinking/ready-fire-aim-harnessing-action-and-ai-for-rapid-progress) (2025-05-27): In an age defined by velocity, the traditional wisdom of "Ready, Aim, Fire" feels increasingly out of sync. Here's how "Ready, Fire, Aim" with AI as your co-pilot enables rapid learning and real-world feedback. - [Why I Don't Niche Down (And Why That's a Strength)](https://leantonio.me/thinking/why-i-dont-niche-down) (2025-04-08): In an industry that values specialisation, being a generalist is often viewed as a weakness. But I've found that connecting disciplines and moving fluidly between design, development, strategy, and execution creates unique value. - [My Process for Turning a Blank Canvas into a Working Prototype in 24 Hours](https://leantonio.me/thinking/24-hour-prototype) (2025-03-12): A look into my process for rapidly moving from a brief to a functional prototype in 24 hours, focusing on clarity, direction, and momentum by reducing friction and using the right tools and mindset. - [Evaluating and Choosing a "Reasonable Extreme" in Technical Creative Ideation](https://leantonio.me/thinking/evaluating-and-choosing-a-reasonable-extreme-in-technical-creative-ideation) (2025-02-14): Discover how creative technologists can identify and select ideas that push boundaries while remaining practical. Explore the balance between innovation and feasibility in technical creative ideation. - [How I Use AI to Build Faster, Think Deeper, and Stay Human](https://leantonio.me/thinking/how-i-use-ai) (2025-01-25): AI is a core part of my workflow, acting as a multiplier for skill, not a replacement. I use it to get to the good stuff faster, remove friction, and enhance creative thinking, all while keeping human judgment at the center. - [What I Look for Before Writing a Line of Code](https://leantonio.me/thinking/what-i-look-for-before-writing-code) (2024-12-10): The most expensive problems in digital projects aren't technical, they're strategic. Before writing code, I focus on clarity, validation, and alignment to ensure we're building the right thing, in the right way, at the right time. - [Personal Reflection: Embracing Change and Failing Fast with AI in Software Development](https://leantonio.me/thinking/embracing-change) (2024-10-30): AI-powered tools have transformed software development by enhancing creativity, streamlining workflows, and fostering a fail-fast approach, allowing developers to adapt quickly and innovate more effectively. - [Crafting Evolution: The Symbiotic Odyssey of Humanity and Technology](https://leantonio.me/thinking/crafting-evolution-the-symbiotic-odyssey-of-humanity-and-technology) (2024-09-15): The journey of human progress is inextricably linked with technological innovation. This article explores the symbiotic relationship between humanity and technology, highlighting how each shapes the other in a continuous cycle of development and adaptation. - [Bridging the Gap: Integrating Empathy into UX Design](https://leantonio.me/thinking/bridging-the-gap-integrating-empathy-into-ux-design) (2024-08-01): Empathy in UX design means understanding user motivations and frustrations to create products that genuinely resonate. It involves active listening, observation, and iteration to build intuitive and supportive digital experiences. - [Designing UX for an AI-Driven Future](https://leantonio.me/thinking/designing-UX-for-an-AI-driven-future) (2024-07-20): Large language models are upending traditional graphical user interface paradigms, ushering in an era where users will naturally converse with intelligent AI assistants through voice and text inputs, requiring user experience designers to radically rethink interaction models for this conversational AI future. - [Digital Transformation: Embracing the Technological Revolution in Business](https://leantonio.me/thinking/embracing-the-technological-revolution-in-business) (2024-06-15): Digital transformation reshapes industries by deeply integrating digital solutions into operations, customer engagement, and culture, focusing on innovation, enhanced customer experience, operational agility, workforce empowerment, and robust cybersecurity to maintain competitiveness in the digital era. The essays are reproduced in full in the appendix. ## 7. Career and work register ### Career [Stated] From the work history on his previous portfolio: - 2010–2012: Hibu - 2012–2014: FitPro - 2014–2016: RadioCentre, senior digital designer - 2016–2017: Tonic, front-end developer - 2017–2020: Tonic, technical lead - 2020–present: independent. Tonic continues as a long-running client; direct clients and ventures since, through Merlin Studio. ### Work, by relationship **Tonic (embedded partnership)** - **Tonic → easyJet careers site**: Tonic's client. Titled Project Director. Fullest account: leantonio.me/projects/easyjet (a fuller Merlin case study does not exist yet). - **Tonic → EY careers sites (UK, US and global)**: Tonic's client. Titled Digital Strategy & UX Lead on his write-up (the same page's header says Project Director; the inconsistency is his page's). Fullest account: leantonio.me/projects/ey. - **Tonic → Baker Hughes careers site**: Tonic's client. By his attested account: audited the existing site, created the wireframes, defined the design direction and coded the site on Phenom, the client's recruitment platform. Fullest account: no public case study yet. - **Tonic → Imperial Brands careers work**: Tonic's client. Titled Project Director on the main build. Fullest account: leantonio.me/projects/imperial. - **Tonic → Havering Council jobs site**: Tonic's client. Project lead and developer. Fullest account: leantonio.me/projects/havering-council. - **Tonic → Lonza small-molecules decision tree**: Tonic's client. Titled Project Director. Fullest account: leantonio.me/projects/lonza. - **Tonic → Open Society Foundations careers site**: Tonic's client. By his attested account: the audit and the sitemap. Fullest account: no public case study yet. - **Tonic → Refinitiv**: Tonic's client. His role is not established. Fullest account: no reliable case study. The later "Refinitiv platform modernisation" page on his old portfolio was not a record of this work and is not used. - **Tonic → Currys**: Tonic's client. Stated: a long-standing arrangement: web build, audits, proposals and pitches, technical direction and support. Fullest account: no public case study. - **Tonic → SSCL**: Tonic's client. Stated: audit, discovery, user journeys, sitemap, wireframes, build and maintenance. Fullest account: no public case study. - **Tonic → Tetra Pak**: Tonic's client. Stated: audit, user journeys, sitemap, wireframes and implementation on Adobe Experience Manager. Fullest account: no public case study. - **Tonic → IHG**: Tonic's client. Stated: the "Room to Grow" internal project (concept, design and build) and a careers-site audit. Fullest account: no public case study. - **Tonic → Kaizen Gaming**: Tonic's client. Stated: careers and corporate site: wireframes, sitemap, user journeys, analytics and the website pitch. Fullest account: no public case study. - **Tonic → Talent Mirror**: work for Tonic itself. Merlin Studio, his studio, built and runs the production software. Fullest account: merlinstudio.io/ventures/talent-mirror. - **Tonic → Tonic website prototype**: work for Tonic itself. Proposed, designed and prototyped it. Fullest account: no public case study yet. **Direct clients** - **Oracle FM website**: direct client. Sole developer: every commit in the repository is his (82, January to October 2026). Fullest account: leantonio.me/projects/oracle-fm. - **The ADHD Clinic for Women website**: direct client. Sole developer: every commit is his. Fullest account: leantonio.me/projects/adhd-clinic. **Ventures and products** - **Scaffolds**: His own product, originated and run through Merlin Studio. Not Tonic's or any client's. Live and on sale at scaffolds.design (free for one project, paid Pro and enterprise tiers). Fullest account: merlinstudio.io/ventures/scaffolds. - **MyEasyTherapy**: A venture partnership, as Merlin Studio describes it: the partner, clinical psychologist Dr Michaela Dunbar, brings the market and the clinical content; Merlin owns everything technical (the app, the sales site, billing, lifecycle email, analytics and the app stores). Live on the web, iOS and Android, by subscription, per Merlin Studio; the repository contains the web app and a Capacitor mobile build. Fullest account: merlinstudio.io/ventures/myeasytherapy. - **Ludoo**: A partner venture, as Merlin Studio describes it: the game's owner brings the game and makes the final product calls; Merlin designs, builds and runs everything else. Playable on the web at app.ludoo.online; mobile builds in testing, not yet in the stores. Fullest account: merlinstudio.io/ventures/ludoo. - **Sapien**: Merlin-originated: Merlin Studio describes building the product end to end. Live on the web; physiotherapist review and app-store releases in progress, per Merlin Studio. Fullest account: merlinstudio.io/ventures/sapien. - **ScratchManager**: His own product, through Merlin Studio; a distribution partner is sought. In private testing on macOS; not on sale, and its own documentation says no build should be called commercially ready until payment, notarisation and signed updates are done. Fullest account: merlinstudio.io/ventures/scratchmanager. - **In the Middle of All Things**: His own publishing product. Live at middleofallthings.com as a reader and listener for his book, with interest registration for the print edition. Fullest account: middleofallthings.com. - **leantonio.me**: His own site. Live. Fullest account: leantonio.me. ## 8. Ways to work together These are the moments people usually bring him in for; read them as where the operating pattern above gets applied, not as separate services. - **Win it.** Pitch support, feasibility, prototypes, technical responses, architecture, estimates and RFP/RFI input. The technical half of a bid, written so it survives the client's own engineers reading it. - **Shape it.** Discovery, research, journeys, IA, wireframes and requirements. Taking an ambiguous brief to the point where somebody can cost it and build it. - **Ship it.** Product leadership and technical direction through delivery, hands-on where that is the fastest way through. - **Fix & grow it.** Audits and optimisation on things that already exist and are not working hard enough. With agencies, the arrangement can be client-facing or silent, and agency-delivered work is never presented as his own client relationship. Modes: Client-facing (In the room, in your brand); Embedded (Inside your team and your process); White-label (Silent. Your name on the work); Pitch only (In for the bid, out after the decision). Ventures and commissioned product work go through [Merlin Studio](https://merlinstudio.io): partners bring a market, Merlin builds the product. Planning tool: [Scaffolds](https://scaffolds.design). Contact: the "Get in touch" form at https://leantonio.me. To talk to an AI version of him, or book a 20-minute call with the real one: https://leantonio.me ## 9. Source and evidence notes **Sources behind this record** - The repositories for Oracle FM, the ADHD Clinic for Women, MyEasyTherapy, Ludoo, ScratchManager, In the Middle of All Things and Tonic's website prototype: code, documentation and commit history. - His previous portfolio, including the git history that separates his first-hand 2025 write-ups from a later bulk rewrite that inflated them. - Merlin Studio's site (his studio): the venture pages and its description of the Tonic relationship. - His own attested record of what he did on each Tonic engagement (testimony, labelled Stated). - His essays (labelled Argument). **Corrections to earlier versions of this record and of his old portfolio.** If you have seen an earlier version, these supersede it. - easyJet: "spearheaded the strategic pivot" is restated from his first-hand account: after consultations with easyJet's head of IT, he moved the project to Contentful and Netlify. - Imperial Brands: "architected the migration strategy" and "perfect responsiveness" came from the later rewrite, not his first-hand account, and are removed; his first-hand account credits Tonic's UX and design teams with research and design. - Refinitiv: the "platform modernisation" page ("rendering millions of data points in real time", and descriptions of Refinitiv's own trading, data and machine-learning products) was not a record of his work and is excluded. His role on Refinitiv is unknown. - Tonic: previously described in this record as his employer throughout; he was employed 2016–2020 and Tonic has been a client since. "I provided the essential digital knowledge and technical skills to bridge this gap" is kept only as his framing of his part. - ADHD Clinic for Women: "zero data loss" is an unsupported assertion and is removed; the site's CMS is now Pages CMS, not Sanity; the clinic's promotional positioning (expert credibility, proprietary tool, audience) is not evidence about him and is removed. - MyEasyTherapy: the old page's therapist-matching algorithm ("custom" or "AI-powered"), HIPAA-compliant video sessions, insurance payments and a sessions count are not present in the product's codebase and are removed. The product is an AI-companion wellness app built as a venture partnership. - Oracle FM: "Relationship: not established" is resolved: a direct client. The "AI-powered tools" are deterministic calculators, and the "AI Compliance Portal" is a concept page with demonstration data; client lists and the client's positioning are removed. - Lonza: the old page said both Angular and React. His first-hand account and every original mention say Angular, chosen by the agency's development team; "React" entered in the later rewrite. - In the Middle of All Things: the public README contained unrelated careers-site boilerplate and describes an earlier version; the old case study also describes that version. Both are excluded; the entry is written from the current repository. - Testimonials on the old portfolio were never verified and are not used. **What the evidence cannot establish** - Kerzner and DHL: he names them among Tonic engagements, but no supporting material was found in the sources reviewed, so they are not presented as evidence. - Outcomes: none is documented for any engagement or product. - Baker Hughes: whether the site has launched. - Sapien and Talent Mirror: their repositories were not reviewed; their entries rest on Merlin Studio's pages. - MyEasyPhilosophy: no evidence found; not included. Everything here is public. For anything else, ask him directly through the "Get in touch" form at https://leantonio.me. ## Appendix: essays in full Arguments, not results. Headings inside essays belong to the essay. ### A Marketer's Guide to Answer Engine Optimization (AEO) https://leantonio.me/thinking/a-marketers-guide-to-answer-engine-optimization · 2025-09-30 The world of search is undergoing a seismic shift. For decades, Search Engine Optimization (SEO) has been a game of ranking on a list of blue links. Today, with the rise of Large Language Models (LLMs) like ChatGPT and Gemini, the game has a new objective: not just to be on the list, but to be the source of the answer itself. Welcome to Answer Engine Optimization (AEO), also known as Generative Engine Optimization (GEO). This is the new frontier of digital marketing, and this guide will walk you through the core principles, new technologies, and actionable strategies you need to succeed. > "The question isn't whether this shift will happen – it's already underway. The real question is: when an AI is asked about your industry, will it recommend you?" ##### The Fundamental Shift: Information vs. Guidance The core difference between traditional SEO and AEO lies in the user's intent and the engine's output. * **Traditional SEO**: You aim to provide the best information for a query, so Google ranks your page highly. * **AEO/LLM SEO**: You aim to provide the best advice, guidance, or recommendation for a prompt, positioning yourself as the authority that an AI trusts and cites. ##### Core Principle: You Must Become the Authority In the world of AEO, everything boils down to one word: **authority**. An LLM's primary goal is to provide a helpful, accurate, and trustworthy answer. It determines this by analysing vast amounts of online data to see which person, brand, or website is most consistently and authoritatively associated with a given topic. Short-term tricks will fade, but a long-term strategy of building genuine authority is the only sustainable way to win. Your goal is to be that undisputed authority. This is achieved through two main pillars: **On-Page Structure** and **Off-Page Prominence**. ##### Pillar 1: On-Page Structure & Content Strategy This pillar is about making the information on your own website perfectly formatted for an AI to understand, process, and reference. ###### 1. Adopt a Question & Answer Content Format Structure your content to directly answer the questions your audience is asking. Use clear, descriptive headings phrased as questions (e.g., "What is the Best Camera for Travel Vlogging?") followed immediately by a concise, direct answer. This direct, parsable format is ideal for AI consumption. ###### 2. Write with Absolute Clarity AI doesn't care about "flashy" or overly creative prose. It values well-structured text with simple, descriptive headlines. Get straight to the point. Use lists, bullet points, and short paragraphs to break down complex topics into digestible chunks of information that can be easily repurposed into an answer. ###### 3. Implement Schema Markup Schema markup is a form of structured data that acts as a translator for search engines and AIs, explicitly telling them what your content is about. For AEO, the most important types are: * **FAQPage Schema**: Perfect for your Q&A pages, as it programmatically labels your questions and answers. * **HowTo Schema**: Use this for step-by-step guides and tutorials. * **Article Schema**: Clearly defines author, publication date, and other metadata, which helps establish the critical signals of Experience, Expertise, Authoritativeness, and Trustworthiness (E-E-A-T). > "Structured content systems are the future. The CMS is evolving into design systems for intent rather than static content stores – this shift directly supports AEO readiness." *Pro tip: Structured content systems are the future.* I've written before about how [the CMS is evolving](https://leantonio.me/articles/why-your-website-needs-to-think) into **design systems for intent** rather than static content stores – this shift directly supports AEO readiness. ###### 4. Focus on Guidance, Not Just Information Your content must have a strong, defensible point of view. It's no longer enough to just present information; you must provide guidance. * Good: A page listing the technical specs of a product. * Better for AEO: A page that compares three products and confidently recommends the best one for a specific use case, clearly explaining the reasoning behind the advice. ###### 5. Optimise for Multimodality LLMs like Gemini are now multimodal. They don't just parse text – they process images, audio, and video. That means transcripts, alt-text, and structured video data are no longer "extras." They're essential inputs for ensuring your brand is part of the training set for visual and voice-based search. ##### Pillar 2: Off-Page Prominence & Seeding the Data An LLM's core understanding comes from the massive datasets it was trained on. Your job is to ensure your brand, expertise, and content are a prominent part of those datasets and the ongoing conversations across the web. ###### 1. Be Active in AI Training Grounds LLMs are heavily trained on data from high-traffic, conversational platforms. You need a presence there, sharing your expertise and linking back to your authoritative content on your website. Key platforms include: * **Reddit**: Find relevant subreddits, answer questions authoritatively, and share your content where it adds value. * **Quora**: As a primary source for Q&A data, this platform is invaluable. * **High-Authority Blogs & Niche Forums**: Participate in the digital communities where your audience and peers are active. ###### 2. Build Brand Mentions and Citable Content The ultimate goal is to have other people and publications cite you as the expert. This is about building brand salience – the degree to which your brand is thought of when a recommendation is needed. * **Publish Original Research**: Create unique data, surveys, or studies that others in your industry will want to cite. * **Develop Strong Opinions**: A well-reasoned, expert opinion is far more likely to be discussed and referenced than a generic summary of a topic. * **Earn Links and Mentions**: When your brand, your name, and your website are mentioned across many trusted public sources in relation to a specific topic, AI models learn to associate you with that expertise. ###### 3. Seed Structured Knowledge Beyond Your Site Knowledge graphs, Wikidata entries, and even structured LinkedIn/author profiles feed into LLM datasets. The more structured and verifiable your digital footprint, the more likely you'll be treated as a reliable authority. ##### Next-Gen Enhancements to Stay Ahead * **RAG & Vector Databases**: Structure your content in a way that can be pulled into retrieval-augmented generation systems, giving you direct pathways into AI-driven answers. See [Orchestrating Intelligence](https://leantonio.me/articles/orchestrating-intelligence-the-new-art-of-software-engineering) for more on this. * **Conversational Layer Optimisation**: Optimise not just for initial queries, but for the follow-up questions an AI user might ask. This is the equivalent of being "sticky" in a chat-based search journey. * **Synthetic Testing with AI**: Don't guess – test. Run your own prompts through different LLMs and track when your brand or advice is cited. Treat this like split-testing for the answer economy. See also [Project Management in the Age of AI](https://leantonio.me/articles/project-management-in-the-age-of-ai). * **AI-Generated Microcontent**: Repurpose your longer content into concise, structured snippets that AIs can easily re-use in answers. ##### The Marketer's New Role Much like I argued in [my writing on agile prototyping](https://leantonio.me/articles/24-hour-prototype), the marketer's job is shifting from static campaigns to dynamic systems. AEO is less about keywords and more about ongoing participation in the datasets that shape generative engines. It's about building lasting authority, seeding structured knowledge, and continuously testing how your brand shows up in the answer economy. > "The question isn't whether this shift will happen – it's already underway. The real question is: when an AI is asked about your industry, will it recommend you?" The future of marketing isn't about being found – it's about being trusted as the source of truth. --- ### Project Management in the Age of AI: Why Sprints and Rapid Prototyping Are Winning https://leantonio.me/thinking/project-management-in-the-age-of-ai · 2025-09-24 For decades, project management has been weighed down by heavyweight methodologies – Gantt charts sprawling like ancient maps, requirements documents fossilised long before a single line of code was written. These approaches worked in an era of predictability, but the present is anything but predictable. We live in a world where AI can rewrite the rules overnight. Yesterday's strategy decks are today's footnotes. The new competitive advantage is not control, but adaptability. ##### Rescuing Agile's Spirit Agile cracked the first fault lines in traditional management by trading rigidity for iteration. But too often it became ceremony – stand-ups without spark, burndown charts without bite. Sprints and rapid prototyping are not a rejection of Agile but a return to its soul: short bets, fast feedback, and working solutions over endless documentation. > "A sprint is a wager. A rapid prototype is proof of that wager." Together, they re-ignite Agile's original intent – moving quickly from idea to impact. ##### What It Looks Like in Practice Picture a product team testing a new onboarding flow for a banking app. On Monday morning, the sprint begins. By lunch, AI has generated five competing wireframes, each tailored to different customer personas. By the afternoon, synthetic user testing has flagged friction points in three of them. By Tuesday, the remaining two flows are live in a sandbox, with real users interacting. By Friday, the team isn't reviewing abstract ideas; they're debating live metrics, engagement rates, and drop-off points. That sprint hasn't just produced a prototype. It has collapsed weeks of design, testing, and iteration into days, with AI as both accelerator and filter. ##### AI Accelerates the Loop AI supercharges this rhythm. It can generate prototypes in minutes, simulate user flows, or test hundreds of variations in parallel. The cost of exploration collapses, making experimentation not only faster but also safer. Yet AI is a moving target – new capabilities appear weekly, making yesterday's roadmaps obsolete. The only way to keep pace is to embrace change as the methodology itself. > "The cost of exploration collapses, making experimentation not only faster but also safer." ##### The Feedback Race Is Already Won This isn't just a new process; it's a survival strategy. Teams stuck in waterfall timelines are outpaced before launch. Teams who sprint, prototype, and adapt inhale uncertainty as oxygen. The winners aren't those with the thickest documentation but those with the fastest feedback loops. ##### Projects as Explorations, Not Expeditions The deeper shift is philosophical. Projects are no longer expeditions from A to B with neatly drawn maps. They're explorations in fog, with visibility stretching only as far as the next sprint. The artefacts that matter most aren't status reports, but living prototypes that breathe, fail, and evolve. > "In the age of AI, the fastest cycle of making and remaking isn't a nice-to-have methodology. It's the only one that survives." Everything else is too slow. --- ### Digital Clarity: Quick Wins SMEs Can Action Today (With AI Prompts You Can Copy and Paste) https://leantonio.me/thinking/digital-clarity-sme-quick-wins · 2025-09-11 #### Digital Clarity: Quick Wins SMEs Can Action Today (With AI Prompts You Can Copy and Paste) Most people who know me know what I do for a living. Over the past few years – especially since Covid – I've noticed a big shift. People have become more entrepreneurial. Friends have launched consultancies, family members have started e-commerce shops, colleagues have turned freelance. And sooner or later, the question comes: *"Can you take a quick look at my website? My SEO? My digital strategy?"* I don't mind – in fact, I enjoy it. But I've noticed the same mistakes appear again and again. The good news? They're fixable. And with the help of AI, even someone with no technical background can start to make expert-level improvements in an afternoon. --- ##### The Biggest Mistakes I See SMEs Making 1. **No analytics** – or set up but never checked. 2. **Slow websites** – oversized images, bloated scripts, no caching. 3. **Confusing homepages** – not clear what you do, who it's for, or how to buy. 4. **SEO as a one-off** – treated like a checklist instead of an ongoing habit. 5. **Fear of AI** – or worse, using it badly (generic content, no context). 6. **Not checking and double-checking** – like a chef continually tasting their food, we must continually test, review, and refine the work we produce. --- ##### Quick Wins You Can Apply Today ###### 1. Set Up Analytics (and Actually Use Them) If you don't measure it, you can't improve it. At the very least, install Google Analytics 4, Search Console, and set up a Google Business Profile. 👉 *AI Boost:* Export your data, paste it into ChatGPT, and ask it to find insights. **Prompt:** ``` I've exported my Google Analytics data. Analyse it and tell me: - Which pages keep visitors the longest - Where most people drop off - What one action I should take to improve engagement ``` --- ###### 2. Audit Your Content for Clarity Your homepage should answer three questions in 10 seconds: * What do you do? * Who is it for? * How do I buy or contact you? 👉 *AI Boost:* Screenshot your homepage and upload it to ChatGPT Vision. **Prompt:** ``` Analyse this screenshot of my homepage. In under 200 words, tell me if it clearly communicates: - What we do - Who it's for - The next step for the visitor Suggest specific improvements. ``` --- ###### 3. Improve Your Website Performance Speed is trust. A slow site turns customers away before you even speak to them. Run your site through Google PageSpeed Insights or GTMetrix. 👉 *AI Boost:* Get technical reports translated into plain English. **Prompt:** ``` Here's my PageSpeed Insights report. Translate the technical recommendations into a plain-English checklist I can give my developer. ``` --- ###### 4. Optimise Your SEO the Right Way Forget shortcuts. For SMEs, SEO is about clarity, consistency, and relevance. Start by writing around the questions your customers actually ask. 👉 *AI Boost:* Use AI to generate content ideas with ready-made meta descriptions. **Prompt:** ``` I run a [type of business]. Suggest 10 blog or page ideas based on questions my customers might search for. Make them SEO-friendly with working titles and meta descriptions. ``` --- ###### 5. Leverage AI at Your Level AI isn't a future thing – it's a right-now tool. SMEs can benefit immediately, depending on their confidence level. * **Beginner**: Rewrite product descriptions, FAQs, or social posts. * **Intermediate**: Automate workflows (e.g. leads from your site go straight into your CRM). * **Advanced**: Train a branded assistant on your own content. 👉 *AI Boost Example (Beginner):* ``` Rewrite this product/service description to make it engaging, clear, and SEO-friendly. Keep it under 150 words. [Insert text here] ``` --- ##### Why This Works Leveraging AI properly at each stage doesn't just save time – it makes you sharper. A novice can become an expert simply by knowing what to ask. Every prompt is a shortcut to insight. Every quick win compounds over time. --- ##### Closing Thought Digital clarity isn't about chasing every new trend. It's about removing clutter, setting up the fundamentals, and layering in smart tools step by step. The difference between SMEs who thrive online and those who vanish isn't budget. It's clarity, consistency, and the courage to try. --- ### Why Your Website Needs to Think: Embracing AI Overviews and AI Mode https://leantonio.me/thinking/why-your-website-needs-to-think · 2025-08-01 > "We're moving from a click-based web to a conversation-based web." > > — Rand Fishkin, SparkToro ##### The Shift: Google's AI Overviews and AI Mode Google now offers two distinct AI-driven search experiences: * **AI Overviews** surface concise summaries at the top of search results. They appeared in **13.14%** of U.S. desktop queries by March 2025, up from **6.49%** in January 2025. Semrush's study of over 200,000 keywords found that zero-click rates for those queries dipped slightly from **38.1%** to **36.2%** once AI Overviews rolled out. > "AI Overviews deliver answers that keep users on the search page, redefining how brands capture attention." > > — Google Search Product Manager * **AI Mode** is a chat-style side panel that tackles complex queries directly. Usage rose from **0.25%** of U.S. desktop sessions at launch to just over **1%** by July 2025, and in **92–94%** of those sessions, users did not click through to any external result. Together, these tools satisfy users with immediate answers and drive a fundamental change in click behavior. ##### Why Traditional Navigation Falls Short Most websites still depend on menus, links and endless scrolling. That design assumes visitors: 1. Know exactly what they want 2. Are willing to navigate several pages 3. Have the time to read long passages In reality, many people now treat content like a conversation: they ask once, get a summary, and decide whether to dive deeper. Static pages and rigid hierarchies introduce friction, lost time, diverted attention and higher bounce rates. > "Zero-click searches now account for almost half of all informational queries." > > — Semrush, March 2025 Study ##### The AI Summary Overlay: A Practical Bridge On [leantonio.me](https://leantonio.me), I've added an **AI summary overlay** to every page. Instead of hunting through menus, visitors are greeted with: * **Instant Summaries** A concise synthesis of key takeaways and practical insights, generated the moment the page loads. * **Contextual Chat** An in-page chat prompt that remembers what's already been summarised and answers follow-up questions on demand. * **Seamless Performance** File-based caching (with content-hash validation and seven-day expiry), a mobile-first UI, and Google Gemini 2.5 Flash for reliable accuracy. This layer transforms static pages into an **interactive knowledge experience**, reducing friction without requiring a complete redesign. ##### Why Recruitment and Brand Sites Should Act Whether you manage a careers portal, product showcase or corporate blog, your audience now expects: * Clear, instant answers * Minimal effort to find relevant information * A conversational, personalised experience An AI summary overlay delivers these benefits on top of your existing site and surfaces real user questions, data you can't capture with clicks and pageviews alone. ##### A Transitional Solution for the AI-Driven Web This overlay isn't the final form of web interaction, but it's a vital first step. Most organisations aren't ready to rebuild from scratch around generative models. Yet they can: * Layer intelligence onto current pages * Track real-time user intent and interests * Introduce conversational UX elements at their own pace By adopting an AI summary overlay today, you position your site for a near future where **dynamic, context-driven interfaces** replace static navigation. ##### A Glimpse Beyond: The Digital Twin Alongside the summary overlay, I've created a **context-aware digital twin** of myself and this site. It draws on every article, project and reflection I've published. Visitors can engage with a version of me that understands both the content and my background, illustrating how personalised AI can deepen engagement in meaningful ways. > This lightweight layer can boost engagement, reduce bounce rates and prepare your digital presence for the evolving expectations of AI-powered search. ##### Next Steps If you'd like to explore how an **AI summary overlay** could transform your site, whether in recruitment, branding or product marketing, let's set up a pilot. This lightweight layer can boost engagement, reduce bounce rates and prepare your digital presence for the evolving expectations of AI-powered search. **Give your audience the intelligence they seek, today.** --- ### Orchestrating Intelligence: The New Art of Software Engineering https://leantonio.me/thinking/orchestrating-intelligence-the-new-art-of-software-engineering · 2025-07-05 > "Software engineering today is less about writing lines of code and more about orchestrating intelligence to solve human problems." In an era when the distinction between human intuition and machine learning grows ever more subtle, true mastery lies not in language syntax but in our capacity to envision how AI can transform lives. As machines become capable of generating boilerplate and even complex routines, our greatest contribution emerges in the form of the questions we choose to ask and the problems we elect to solve. With AI models serving as raw, malleable materials, today's engineer resembles an alchemist, blending data, prompts and system architecture to distil insight and value. > "When machines can write code, our value emerges in the questions we choose to tackle." No longer is proficiency in a single framework or pattern the measure of worth. Instead, what endures is the ability to discern patterns in seemingly chaotic data, to anticipate user needs before they are spoken, and to craft narratives that guide AI toward outcomes that resonate on a human level. Creativity has become our prime constraint. In a landscape of low-code platforms and AI assistants, the question "Can we build X?" fades into irrelevance; the more pertinent inquiries are "Should we build X?" and "How can we build it so elegantly that it feels intuitive and humane?" Engineers must now wear multiple hats, domain expert, ethicist, storyteller, ensuring that every system is built with privacy, fairness and sustainability at its core. As I speculate, the next frontier may see "meta-engineers" orchestrating ensembles of specialised AI agents, some fine-tuned for legal reasoning, others for user-experience critique, all collaborating in a seamless symphony to tackle complex, evolving challenges. Resilience becomes as vital as innovation. The probabilistic nature of AI demands rigorous observability: we design fallbacks for when models hallucinate, and we build rapid-recovery pipelines to maintain trust and reliability. Embracing uncertainty, we establish continuous-feedback loops that allow systems to learn and adapt in production, rather than hoping for perfection at launch. Education, too, must evolve, shifting from rote syntax drills toward immersive workshops on problem framing, ethical AI design and systems thinking. Organisations reflect this shift by dissolving silos: engineers, product managers and domain specialists converge in shared spaces, co-authoring solutions in real time. Impact is measured not in thousands of lines of code, but in emergent value, how gracefully systems empower users, streamline workflows and unlock new possibilities. We become curators of experience, sculpting the interface between human aspiration and machine capability so that every interaction delights as much as it delivers. > "Ultimately, the destiny of software engineering is to mature from a craft of instruction-writing into the art of shaping intelligence." As the instruments of creation grow ever more powerful, we shoulder greater responsibility: to ask not only what we can automate, but what we ought to automate. In that shift, from builder to conductor, from coder to creative technologist, lies the future of our discipline, one where imagination trumps syntax and vision guides every line of code. --- ### Designing from the Middle: A Philosophy of Presence in Product and Code https://leantonio.me/thinking/designing-from-the-middle · 2025-07-01 When people ask about my approach to design and development, I often say I work from the middle. It's not a slogan, it's the most accurate way I can describe how I make decisions. The middle is a point of presence, a place where competing forces can be held together without collapsing into extremes. I try to hold space for both the user's immediate need and their longer journey. I'm not interested in building products that dazzle or direct. The goal is to meet people where they are, offer quiet clarity, and evolve with them. A good product doesn't try to impress, it allows the user to recognise something of themselves within it. The same principle guides my technical decisions. Development isn't something to rush. It's something to hold with care. I don't swing between extremes, between rapid experimentation and long-term stability, or simplicity and abstraction. I stay with the tension. I ask: can this system flex without breaking? Does it serve the moment while preparing for the future? Every decision is a balancing act between what we know and what we're stepping into. And in that tension, if we don't rush it, resilience begins to emerge. When it comes to shipping, I don't ask if a product is finished, I ask if it's centred. Completion isn't something to chase. It's something to notice. A product is ready when it holds enough coherence to speak for itself. I don't ship based on pressure; I ship when the release feels like an exhale. Balanced, not perfect. Grounded, not rushed. There's a perceived tension between user needs and business goals, but I don't see them as opposites. The user brings energy; the business provides direction. Neither should dominate. My role is to find the middle path where empathy meets strategy. Alignment isn't something to force. It's something to reveal through honest listening and intentional decisions. Design, to me, isn't something to impose. It's something to translate. The interface shouldn't push the user around or hold their hand too tightly. It should feel like a mirror, not a maze. The job of design is to reveal what already wants to emerge, through clarity, space, and quiet symbolism. The best design doesn't demand attention. It rewards awareness. I've learned not to fight complexity. Complexity isn't something to avoid. It's something to map. When a system feels dense, it usually means something essential is trying to take shape. I slow down and stay with it. I look for the principle that can unify without reducing. Simplicity, if it appears, is often what remains after we've honoured the full shape of the complexity. When prioritising features, I don't just listen to what's loud, I listen to what's resonant. Urgency isn't something to obey. It's something to examine. I look for points in the user's experience where they hesitate, repeat, or invent their own solutions. Those are the signals. The most valuable features often grow out of subtle, consistent tension, not noise. Innovation isn't something to chase. It's something to uncover. When you pause long enough, when you listen deeply to friction, silence, and emergence, you begin to see what's missing. The best ideas don't arrive as breakthroughs. They arrive as things that feel inevitable once noticed. Innovation doesn't need spectacle. It needs space. What makes a product meaningful is its ability to support transformation without pushing for it. It becomes part of the user's rhythm. It doesn't insert itself into their life, it quietly shapes it. Meaning isn't something to design directly. It's something to create space for. The best tools eventually disappear into the background, becoming indistinguishable from the user's own flow. Working with teams is another place where I try to stay in the middle. My job isn't to control or convince. It's to hold a clear centre of intent so that different roles, engineering, design, strategy, can move around it without friction. Alignment isn't something to enforce. It's something to feel. If people are orbiting the same centre, collaboration happens naturally. When it comes to user research, I don't just ask questions, I watch rituals. What do users repeat? What do they avoid? Where do they pause? Often the most useful insight isn't what's said, it's what's skipped. Research isn't something to extract. It's something to witness. The product and the user are already in conversation. The work is to learn how to listen. Failure, like everything else, has a centre. It isn't something to fear. It's something to integrate. I don't treat failure as a problem to fix, I treat it as an adjustment to fold in. Every misstep teaches you where your structures are too rigid or too loose. I build systems that allow for return: short feedback loops, forgiving interfaces, and the freedom to get it wrong without punishment. This is what it means to design, develop, and manage products from the middle. Not to compromise, but to notice. Not to dominate, but to hold. Not to finish, but to reveal. It's about building systems that stay alive, flexible enough to evolve, centred enough to stand. --- ##### Practical Steps by Discipline ###### For Designers: - Use progressive disclosure to invite users into depth without overwhelming them. - Let whitespace carry emotional weight; trust in spaciousness. - Create symbolic visual languages that reflect user states or values. - Prioritise mirroring user intent over directing them. ###### For Developers: - Architect systems for change, not permanence, modularity over monolith. - Balance technical elegance with human use cases; prioritise readability. - Stay present with edge cases, your design lives or dies in the margins. - Use guardrails, not gates, build systems that can fail gracefully. ###### For Product Managers: - Define product readiness based on coherence, not completeness. - Use feedback loops to sense emergent needs before they're articulated. - Frame priorities by user tension and behavioural insight, not just KPIs. - Act as a centre, not a funnel, create shared meaning, not just task lists. --- ### Generative UI: From Static Interfaces to Living Experiences https://leantonio.me/thinking/generative-ui-from-static-interfaces-to-living-experiences · 2025-06-26 Imagine opening a website that doesn't just respond to your clicks, it responds to *you*. Not the generic you of market segments and demographic profiles, but the specific you, in this moment, with your particular needs, context, and emotional state. This isn't science fiction. It's the emerging reality of generative UI, where artificial intelligence doesn't just power recommendations or chatbots, but crafts the very interface you interact with, in real time. We stand at a threshold. Traditional user interfaces are essentially digital fossils, fixed arrangements of buttons, menus, and layouts, designed once and deployed everywhere. But what if interfaces could be living things? What if they could learn, adapt, and evolve with each interaction, becoming more intuitive and personally relevant over time? This is the promise of generative UI: interfaces that generate themselves. ##### From Static to Dynamic: A Fundamental Shift To understand this shift, think about how we've always designed digital experiences. A designer creates wireframes and mockups. A developer builds components. A product team defines user flows. The interface is essentially "frozen" at launch, the same homepage greets millions of users, the same checkout flow processes thousands of orders, the same dashboard presents identical data layouts to executives and interns alike. Generative UI inverts this relationship. Instead of designing static screens, we design systems that design themselves. We create component libraries that an AI can recombine. We establish design principles that guide machine creativity. We feed real-time data, user behavior, device context, environmental factors, even emotional cues, into intelligent systems that compose interfaces on demand. Picture this: You're browsing an e-commerce site at 11 PM on a Sunday, having just searched for "comfortable work shoes." The AI notices your late-night browsing pattern, your search history, your previous purchases, even your location's weather forecast. Instead of showing the generic homepage, it generates a personalized interface: softer colors for evening viewing, prominent comfortable shoe recommendations, a simplified checkout process recognizing your purchase intent, and perhaps even next-day delivery options knowing Monday is a workday. This isn't just smart personalization, it's the interface itself being born anew for every interaction. ##### The Architecture of Adaptive Interfaces Generative UI rests on three foundational elements, working in concert like instruments in an orchestra: **The Component Library**: Traditional UI components (buttons, cards, headers, forms) become "smart" building blocks. Each component is designed with multiple variations and the ability to adapt its visual style, content, and behavior based on context. Think of these as Lego blocks that an AI architect can arrange into countless configurations. **The Data Stream**: Real-time information flows continuously into the system, analytics showing how users interact with different elements, A/B test results revealing what resonates, CRM data indicating customer preferences, device capabilities, time of day, location, even biometric feedback from wearables. This data becomes the raw material for interface decisions. **The AI Orchestrator**: The intelligence layer that interprets all this data against design principles, brand guidelines, and user goals. It's like having a brilliant designer who never sleeps, constantly analyzing patterns and crafting personalized experiences at the speed of milliseconds. When someone visits your site, the orchestrator instantly processes their context: "First-time visitor from mobile device, 2 PM on Tuesday, referred from social media, interested in premium products based on browsing pattern." In real-time, it assembles an interface optimized for this specific moment, perhaps a mobile-first layout with social proof elements, premium product highlights, and a streamlined path to high-value actions. ##### The Learning Loop: Interfaces That Evolve Here's where generative UI becomes truly powerful, it learns from every interaction. Traditional interfaces improve through lengthy design cycles: observe user behavior, hypothesize improvements, design new versions, test, and deploy. This process takes weeks or months. Generative UI compresses this cycle into real-time. The moment someone interacts with a generated interface, clicking a button, scrolling past content, spending time on specific sections, that data feeds back into the system. The AI adjusts not just for that individual user, but incorporates learnings across all users, constantly refining its understanding of what works. This creates a form of accelerated evolution. Interfaces become living experiments, continuously testing new combinations of elements, layouts, and interactions. Poor-performing variations naturally fade away. Successful patterns spread and evolve. Over time, the system develops an increasingly sophisticated understanding of how to create compelling experiences for different contexts and user types. It's reminiscent of biological evolution, but operating at digital speed, thousands of "generations" of interface improvements happening daily rather than over millennia. ##### Redefining Creative Roles This transformation demands new ways of thinking about design and development roles. Designers shift from crafting pixel-perfect mockups to creating "design DNA", establishing the visual language, interaction principles, and emotional tone that the AI will express through countless variations. They become conductors rather than painters, setting the rhythm and melody while allowing for improvisation. Developers build not just applications, but creative systems, component libraries that can recombine gracefully, APIs that provide rich contextual data, and constraints that ensure generated interfaces remain accessible, performant, and on-brand. Product managers evolve from writing detailed specifications to setting strategic parameters: "Prioritize trust-building for first-time visitors," "Emphasize sustainability messaging for environmentally conscious users," "Simplify decision-making for users showing analysis paralysis." Most intriguingly, users themselves become co-creators. Their preferences, behaviors, and reactions directly shape future interface generations. In advanced implementations, users might even provide explicit guidance: "I prefer visual content over text," "Show me technical details upfront," "I'm easily overwhelmed by too many options." ##### The Emotional Dimension The most profound potential of generative UI lies not in its technical capabilities, but in its capacity for emotional resonance. Current interfaces are emotionally flat, they present the same cheerful animations to a frustrated customer and a delighted one, the same urgent call-to-action to someone ready to buy and someone just browsing. Generative UI opens possibilities for emotional intelligence at scale. By analyzing interaction patterns, voice tone in customer service calls, text sentiment in feedback forms, even facial expressions in video interactions, interfaces could adapt their emotional tenor in real-time. A user struggling with a complex task might see gentler transitions, more supportive microcopy, and clearer visual hierarchies. Someone expressing frustration could be presented with calmer colors and more direct paths to human help. A confident, experienced user might see more advanced options and faster interactions. This isn't about manipulation, it's about matching the interface's emotional frequency to the user's current state, creating more harmonious and effective interactions. ##### Navigating the Challenges Like any powerful technology, generative UI brings important considerations. There's the risk of creating "filter bubbles" where users only see interfaces that confirm their existing preferences, potentially limiting discovery and growth. Privacy concerns become more complex when systems require deeper behavioral data to function effectively. Quality control presents unique challenges, how do you ensure brand consistency across millions of generated variations? How do you prevent AI from creating inappropriate or exclusionary experiences? These require sophisticated guardrails, continuous monitoring, and clear ethical frameworks. There's also the fundamental design challenge of maintaining coherence and usability while embracing variation. Too little adaptation and you lose the benefits; too much and you risk confusing users with constantly shifting interfaces. ##### Building Toward Tomorrow For organizations considering generative UI, the path forward is evolutionary, not revolutionary. Start with small experiments, dynamic content areas, adaptive navigation elements, or personalized onboarding flows. Build the data infrastructure to capture rich user signals. Develop component libraries designed for flexibility and recombination. Most importantly, maintain focus on human needs rather than technological capabilities. The goal isn't to showcase AI sophistication, it's to create more intuitive, helpful, and delightful experiences for real people with real goals. As we move forward, successful implementations will likely follow familiar patterns of technological adoption: starting with simple, controlled applications and gradually expanding as we develop better understanding of best practices, user expectations, and ethical frameworks. ##### The Future of Interface Design Generative UI represents more than a new design technique, it's a fundamental shift toward interfaces that can truly understand and adapt to human complexity. We're moving from one-size-fits-all to perfectly-fitted, from static presentations to dynamic conversations, from designed experiences to co-created ones. This isn't about replacing human creativity with machine generation. It's about amplifying human insight through intelligent systems, creating interfaces that can scale empathy, understanding, and helpfulness across millions of interactions. The next time you encounter a frustratingly generic interface, a checkout flow that doesn't understand your urgency, a dashboard that buries the information you actually need, a form that asks for details it should already know, imagine instead an interface that understands your context, adapts to your preferences, and evolves based on your feedback. That's the future generative UI promises: digital experiences as unique, responsive, and ever-improving as the humans who use them. --- ### Ready, Fire, Aim: Harnessing Action and AI for Rapid Progress https://leantonio.me/thinking/ready-fire-aim-harnessing-action-and-ai-for-rapid-progress · 2025-05-27 In an age defined by velocity, where opportunities slip by in the time it takes to draft a plan, the traditional wisdom of "Ready, Aim, Fire" feels increasingly out of sync with the real world. For decades, the sequence offered a sense of control: gather your information, study your target, adjust for the wind, and only then pull the trigger. This process, while sound in theory, all too often becomes a waiting game, paralysing us in the no-man's-land between preparation and action. In many creative and entrepreneurial spaces, the price of hesitation is simply too high. The world changes while you plan. Windows close, competitors move, inspiration fades. Here is where the principle of "Ready, Fire, Aim" asserts itself. At first glance, it seems almost anarchic in its disregard for the sanctity of perfect planning. But on closer inspection, this approach is not about recklessness; it is a philosophy built on the foundation of forward momentum, learning through doing, and letting reality, rather than theory, be your compass. It asks: how much do you actually need to know before you begin? The answer, more often than not, is "just enough." ##### AI as the Ultimate Companion With the arrival of artificial intelligence, this mindset has found its ultimate companion. Where once the cost of experimentation was high, requiring time, money, and risk, AI compresses the space between idea and execution. Now, you can gather insight in moments. You can spin up prototypes, generate copy, analyse markets, or even draft entire business models with a handful of prompts. AI is not a replacement for judgment, but an accelerator for it, enabling you to "get ready" faster than ever before. The true leap happens when you decide to fire. You take your early, imperfect version, a minimum viable product, a sketch, a proposal, a new habit and you release it into the wild. You watch what happens. This is where theory and reality part ways. In my work, I regularly use AI to cross this threshold. Rather than spending weeks refining an idea in a vacuum, I'll use AI tools to create a working version in a day, sometimes even an hour. I might use AI to draft a pitch, design a landing page, simulate a conversation with a prospective user, or even critique my own code. The cost of these experiments is low, often just time and attention. And because I am willing to act before I am entirely certain, I get feedback that is more honest, direct, and actionable than anything that emerges from speculation. ##### The Learning Cycle But "firing" is not the end of the process, it is the beginning of a much richer cycle. Once you've put something into the world, you have the raw material for genuine learning. Here again, AI becomes indispensable. I use it to analyse reactions, sift through feedback, cluster comments, and surface patterns invisible to the naked eye. I can simulate different outcomes, test variations, and even forecast how a new direction might play out, all before making the next adjustment. In this way, aiming is not just about fine-tuning your shot, but understanding the very shape of the landscape you're moving through. Each iteration sharpens your vision and increases your confidence, not because you waited for perfect clarity, but because you acted in its absence. This approach does not invite chaos. It invites a partnership between intuition and intelligence, between boldness and reflection. It acknowledges that most of us overestimate the risk of moving too early and underestimate the risk of moving too late. In highly regulated or safety-critical domains, of course, the old wisdom still holds. But for anyone building, writing, designing, or inventing, speed of learning is now the sharpest edge. ##### Redefining Readiness "Ready, Fire, Aim" is not about dismissing preparation, but about redefining what it means to be ready. Sometimes the best way to learn where the target is, or even what the target should be, is to launch your arrow and see where it lands. The world will give you feedback that no amount of theorising can supply. With AI as your co-pilot, you can try more things, at lower cost, with smarter feedback, and adapt in real time. Every "miss" is simply new data, new fuel for your next attempt. So if you find yourself stalling, waiting for that elusive moment of perfect confidence, ask yourself: is your caution truly protecting you, or just holding you back from the cycle of action and improvement? In a reality shaped by rapid change and constant feedback, mastery belongs to those who move, learn, and recalibrate in rhythm with the world. Fire, then aim, and let the wisdom of the machine, combined with your own, chart the course as you go. --- ### Why I Don't Niche Down (And Why That's a Strength) https://leantonio.me/thinking/why-i-dont-niche-down · 2025-04-08 The question comes up often: "What's your niche?" > "In an industry that values specialisation, the assumption is that you should focus on one thing and build your reputation around it. But in my case, the opposite has been true." I started my career as a web designer. From there, I moved into digital and product design, then into front-end development. I've worked as a technical lead and served as a technical director, managing teams and delivering full-scale digital solutions across a range of industries. Each of these roles required deep expertise, and I didn't skip steps. I built skills from the ground up, layer by layer. > "The true value I bring isn't limited to a single role. It's the ability to connect disciplines, to move fluidly between design, development, strategy, and execution." I'm not just a generalist by preference, I'm a generalist by evolution. ##### What That Means in Practice I've worked across the full stack. I've built apps, architected systems, designed user journeys, led teams, managed clients, and delivered high-performance products. I'm equally comfortable writing production code, guiding AI development, or scoping out the best way to meet business goals. My approach is pragmatic: I prioritise what works, what scales, and what makes sense for the people using it. > "The most impactful solutions often come from seeing across layers and connecting the dots in ways that specialists can't always do alone." When needed, I go deep. But I've learned that the most impactful solutions often come from seeing across layers, business, user experience, data, design, and engineering, and connecting the dots in ways that specialists can't always do alone. ##### Who I Work Best With Clients and collaborators who value versatility, strategic thinking, and clean execution. People who need someone who can wear multiple hats without losing focus. I work well with early-stage startups, scaling teams, or organisations looking to build something unconventional. If you're looking for a partner who brings a broad skill set, a strong sense of ownership, and a track record of delivering real outcomes, I'd be happy to talk. --- ### My Process for Turning a Blank Canvas into a Working Prototype in 24 Hours https://leantonio.me/thinking/24-hour-prototype · 2025-03-12 Not every project needs a sprint team, three rounds of stakeholder alignment, and a six-week timeline. Sometimes, the goal is clarity. Direction. Momentum. That's where I thrive. > "It's not about rushing, it's about reducing friction." Over the years, I've developed a process that lets me turn a blank brief into a functional prototype in 24 hours or less. Using the right tools, the right mindset, and a tight loop between idea and execution. ##### Here's How It Works ###### 1. Clarify the Core Idea Everything starts with one sharp question: What are we trying to prove, test, or feel? In this first phase, I'm not building, I'm reducing. Stripping the idea down to its essentials. Who's the user? What do they need to understand or do? What's the moment of value? This step takes an hour, max. No slides. Just focused conversation and notes. ###### 2. Set Boundaries Scope is everything. I define clear limits: - What must be included? - What can be stubbed or mocked? - What can we ignore, for now? > "Constraints sharpen creativity. Without them, speed becomes noise." ###### 3. Use AI & Boilerplates to Scaffold Once direction is clear, I bring in my tools: custom boilerplates, component libraries, and AI-assisted scaffolding. Whether it's Next.js, Firebase, Tailwind, or a headless CMS, I start with pre-built patterns that I trust and can move quickly with. AI helps accelerate structure, layouts, schema, form logic, naming conventions, so I can focus on the hard parts. ###### 4. Build the Interaction Layer First I always begin with what the user feels. Even if the backend is mocked, even if data is static, users should be able to click, move, flow through something that feels real. Because once they feel it, they can respond to it. From there, I layer in real logic piece by piece. ###### 5. Polish Just Enough At this stage, it's about clarity, not perfection. Design is clean but minimal. Copy is intentional but draft-level. The point is to create something functional enough to demo, test, or pitch, without wasting time on pixel-pushing. ##### Why This Works > "The 24-hour prototype isn't a final product, it's a decision-making tool. Something to see, touch, and talk about." > > — Leantonio Nelson This process works because it's built on experience. I've already worked as the designer, developer, strategist, and director. I know where time is often wasted, and where it's worth leaning in. And sometimes, that's exactly what moves a project forward. If you need to validate a concept, align stakeholders, or explore direction without months of build time, I can help you move from idea to interface in a single day. --- ### Evaluating and Choosing a "Reasonable Extreme" in Technical Creative Ideation https://leantonio.me/thinking/evaluating-and-choosing-a-reasonable-extreme-in-technical-creative-ideation · 2025-02-14 In the realm of creative technology, the pursuit of innovation demands a delicate balance between unrestrained imagination and practical feasibility. As a creative technologist, navigating this balance requires the ability to evaluate and choose a "reasonable extreme", an idea that stretches boundaries while remaining achievable. This article delves into the process of identifying and selecting such ideas during brainstorming and ideation. **The Role of Extremes in Creativity** Extremes play a pivotal role in creative thinking. They serve as catalysts for breaking conventional patterns and fostering unique solutions. However, not all extreme ideas are practical or beneficial. The key is to discern which extreme ideas hold potential for real-world application without compromising on creativity or technical feasibility. **The Brainstorming Phase** Effective brainstorming is the foundation for discovering extreme ideas. During this phase, it's crucial to foster an environment that encourages free thinking and the exploration of unconventional concepts. Techniques such as mind mapping, SCAMPER (Substitute, Combine, Adapt, Modify, Put to another use, Eliminate, Reverse), and role-storming can help in generating a wide range of ideas. 1. **Mind Mapping**: Visualising connections between ideas helps in identifying relationships and potential areas for innovation. 2. **SCAMPER**: This technique stimulates creative thinking by prompting questions that lead to new perspectives. 3. **Role-Storming**: Assuming different personas can uncover insights and ideas that might not surface in a traditional brainstorming session. **Evaluating Ideas: Criteria for Reasonable Extremes** Once a plethora of ideas is on the table, the evaluation phase begins. This is where the concept of "reasonable extremes" comes into play. Evaluating these ideas involves several criteria: 1. **Feasibility**: Assess whether the idea can be realistically implemented with current or near-future technology. Consider technical constraints, resource availability, and time frames. 2. **Impact**: Evaluate the potential impact of the idea. Does it solve a significant problem or create substantial value? The idea should promise a meaningful improvement or innovation. 3. **Scalability**: Consider whether the idea can be scaled effectively. An idea that works on a small scale but fails to grow with demand might not be sustainable. 4. **Alignment with Goals**: Ensure the idea aligns with the broader goals and objectives of the project or organisation. Even the most innovative concept should contribute to the overarching vision. 5. **Risk vs Reward**: Analyse the risks associated with the idea against the potential rewards. While high-risk ideas can yield high rewards, it's essential to balance ambition with prudence. **Techniques for Choosing the Right Extreme** 1. **SWOT Analysis**: Conducting a SWOT (Strengths, Weaknesses, Opportunities, Threats) analysis helps in understanding the internal and external factors that could influence the success of an idea. 2. **Prototyping**: Building prototypes or MVPs (Minimum Viable Products) allows for testing and refining ideas in a practical context. This step is crucial for identifying unforeseen challenges and gauging real-world feasibility. 3. **Feedback Loops**: Engaging with stakeholders, including team members, potential users, and industry experts, provides diverse perspectives that can highlight potential pitfalls or enhancements. 4. **Iterative Refinement**: The iterative process of refining ideas based on feedback and testing ensures continuous improvement and adaptation to changing conditions. **Case Studies: Examples of Reasonable Extremes** 1. **Google Glass**: Initially conceived as a groundbreaking wearable technology, Google Glass represented an extreme in augmented reality. While the first iteration faced challenges, the concept has evolved, finding niche applications in industries like healthcare and logistics. 2. **Tesla's Autopilot**: Pushing the boundaries of autonomous driving, Tesla's Autopilot system exemplifies a reasonable extreme. By incrementally advancing the technology and incorporating extensive user feedback, Tesla continues to innovate within a challenging regulatory and technical landscape. 3. **3D Printing in Medicine**: The use of 3D printing to create customised medical implants and prosthetics showcases an extreme idea made reasonable through advances in material science and printing technology. The impact on personalised medicine and patient outcomes is profound. **Conclusion: Embracing Reasonable Extremes** For a creative technologist, the journey from ideation to implementation is marked by the quest for reasonable extremes. By fostering an environment conducive to radical thinking, rigorously evaluating ideas, and iteratively refining concepts, it is possible to achieve innovation that is both visionary and practical. Balancing creativity with technical feasibility ensures that the chosen extremes are not only bold but also capable of making a tangible impact in the real world. In the end, the art of selecting a reasonable extreme lies in recognising the sweet spot where imagination meets reality, paving the way for groundbreaking advancements that push the boundaries of what's possible. --- ### How I Use AI to Build Faster, Think Deeper, and Stay Human https://leantonio.me/thinking/how-i-use-ai · 2025-01-25 AI has become a core part of how I work, not as a replacement for skill, but as a multiplier. I don't use it to cut corners. I use it to get to the good stuff faster. To remove friction. To move from idea to execution with less waste and more clarity. But beyond speed, AI has also shifted how I think. It's become a creative partner, one that helps me explore more possibilities without losing focus on what matters most: the people we're building for. > "AI is a multiplier for skill, not a replacement. I use it to get to the good stuff faster." ##### Faster Doesn't Mean Rushed When I say AI makes me faster, I don't mean I'm automating the entire build. I still design, develop, and debug with intent. What AI helps me do is reduce cognitive load, generate boilerplate, suggest component structures, scaffold features, or summarise dense technical docs. That means I spend less time on setup and more time refining, solving, and shipping. It's not about shortcuts. It's about streamlining the parts that don't need to be slow. ##### Thinking With, Not Just Through I also use AI to sharpen my thinking. I treat it like a second brain, a sounding board for architecture decisions, product logic, or even brand voice. I prompt it with constraints, contradictions, or half-formed ideas and use its responses to stress-test my assumptions. Sometimes, it shows me a new way in. Other times, it confirms what I already know. Either way, the process is collaborative. ##### Staying Human in the Loop What makes the work good isn't the tech, it's the judgment. AI can suggest a layout, but only a human can sense whether it feels right. It can write code, but it can't feel friction in a user journey. It can spot a bug, but not explain the context that gave rise to it. So I stay in the loop. I make the decisions. I read between the lines. > "What makes the work good isn't the tech, it's the judgment." AI gives me more space to do the work that requires care, experience, and human insight. ##### How I Integrate AI Into Projects - Scaffolding early builds – Getting from idea to prototype faster using code generation and layout tools. - Automating repetitive tasks – Like content formatting, test writing, or schema generation. - Enhancing UX workflows – Using AI for suggestions, smart defaults, or intelligent search. - Improving communication – Summarising stakeholder input, writing documentation, or transforming technical language for non-technical teams. - Experimenting creatively – Building agents, voice interfaces, or exploratory tools that push beyond the standard web experience. ##### Final Thoughts > "Used well, AI gives me more time to do the real work: thinking clearly, designing purposefully, and solving meaningful problems." > > — Leantonio Nelson It's not about working less. It's about removing friction between insight and impact. If you're building something and want to move fast without breaking clarity, this is where I do my best work. --- ### What I Look for Before Writing a Line of Code https://leantonio.me/thinking/what-i-look-for-before-writing-code · 2024-12-10 Before I open a code editor, I ask a series of questions, not just about what I'm building, but why, for whom, and how it fits into the bigger picture. In my experience, the most expensive problems in digital projects aren't technical, they're strategic. They come from building the wrong thing, in the wrong way, at the wrong time. > "The most expensive problems in digital projects aren't technical, they're strategic." That's why I treat planning, scoping, and early alignment as part of the build process, not as a separate phase. Code is execution. But clarity is where the real value begins. ##### 1. Is the problem clearly defined? Too many projects start with a solution in mind before the problem has been fully understood. I look for clarity on what we're solving, and why it matters. This includes understanding the user's pain points, the business objectives, and the desired outcomes, not just features. > "If we can't explain the core problem in one sentence, we're not ready to build." ##### 2. Does the solution align with real-world use? I always aim to pressure-test ideas before they're coded. That might mean using Figma for a quick prototype, running a stakeholder workshop, or just asking hard questions. Will users actually do this? Does this reduce friction or introduce it? Is there a simpler version worth testing first? My goal is to cut through assumptions and reduce waste. ##### 3. What's the fastest way to validate it? I'm a big believer in building lean, even in complex systems. I look for the quickest, cleanest path to a working version, one that gives us feedback early. That might involve reusable components, boilerplates, AI-assisted scaffolding, or integrating existing tools instead of reinventing the wheel. Speed doesn't mean rushing. It means reducing unnecessary decisions. ##### 4. How will this scale and evolve? Even when building MVPs, I think about what comes next. Can this architecture support new features? Will this tech stack hold up under load? Will we regret this decision in six months? I aim to avoid over-engineering, but I also build with modularity and change in mind. Projects grow. The codebase should be ready for that. ##### 5. Who needs to understand this later? I write and organise code so that it's understandable not just to me, but to the next person who works on it. That means clear naming, purposeful comments, and a structure that makes sense. I also consider the documentation, onboarding, and knowledge-sharing practices that surround the code. Because sustainable systems aren't just built, they're maintained. > "Writing code is the easy part. Writing the right code, the part that solves the right problem, integrates well, scales properly, and makes people's lives easier, that's where the real work is." And that work starts long before the first line is written. If you're looking for someone who thinks before building, challenges assumptions, and delivers with intention, I'd be happy to connect. --- ### Personal Reflection: Embracing Change and Failing Fast with AI in Software Development https://leantonio.me/thinking/embracing-change · 2024-10-30 In my journey as a developer, integrating AI into my workflow has brought about significant transformation. The willingness to change has been a driving force behind this evolution. AI-powered coding assistants have not only streamlined my development process but also unlocked a level of creativity I hadn't accessed before. Previously, turning a creative idea into a tangible product often felt slow and frustrating. While my front-end skills were solid, I frequently had to rely on others for backend expertise or spent hours debugging code. This hampered my creative flow and delayed the realisation of innovative concepts. With the advent of AI tools like GitHub Copilot and OpenAI Codex, everything changed. These tools removed repetitive tasks and technical bottlenecks, allowing me to focus on higher-level problem-solving. The AI doesn't replace me as a developer, it augments my capabilities, acting as a collaborative partner in the creative process. Adopting a fail-fast approach has been key to maximising the benefits of AI. In software development, quickly iterating and pivoting is crucial. AI-assisted coding accelerates this process by providing immediate feedback and solutions, enabling me to test ideas rapidly, learn from failures, and refine solutions without wasting time. Failing fast isn't about embracing failure for its own sake; it's about accelerating the learning curve. This approach aligns with the broader philosophy that "the will to change is the key to life and longevity." In both tech and life, resisting change leads to stagnation. By staying open to new methodologies, I've kept my skills relevant and embraced innovation. AI in software development is not just a trend; it's a paradigm shift in problem-solving. A recent project highlights this shift: integrating a complex backend system with a user-friendly front-end interface. Before, I would have needed backend specialists or spent a significant amount of time in unfamiliar territory. With AI, I generated the necessary backend code snippets, understood them through AI explanations, and seamlessly integrated them into my project. This autonomy maintained creative momentum and resulted in a cohesive development process. Embracing change and leveraging AI has transformed my development experience. With a willingness to adapt and a strategy to fail fast, I've enhanced both my productivity and the quality of my work. AI tools empower developers to turn creativity into reality, demonstrating that growth comes from embracing change and learning swiftly from failures. --- ### Crafting Evolution: The Symbiotic Odyssey of Humanity and Technology https://leantonio.me/thinking/crafting-evolution-the-symbiotic-odyssey-of-humanity-and-technology · 2024-09-15 The unbreakable bond between humanity and technology is a narrative etched deep in the grooves of our collective history. As creators of tools and tales, humans have long since ceased to be mere participants in the story of evolution, we've become its authors, penning each chapter with the ink of innovation. Consider the sharpened edge of the first stone tool, an innovation that cut through the fabric of prehistoric life, carving a path towards modernity. This wasn't just a piece of technology; it was the embodiment of a cognitive leap. The toolmakers weren't just surviving; they were learning the language of problem-solving, a skill that would become the foundation of civilization. The agrarian revolution painted another stroke on this canvas. It was a masterstroke that redefined our relationship with the Earth and with each other. Here, technology, rudimentary as it may have seemed, was a force of unity and division, gathering communities around seeded fields while laying the early groundwork for the social hierarchies to come. Leap ahead to the Gutenberg press, a pivot where words transcended the barriers of geography and privilege. The technology of print didn't merely spread information; it democratized knowledge, becoming the soil from which the tree of enlightenment would grow. And now, the digital age. Here, technology merges with the sinew of human experience. Code is the new DNA, and pixels form the windows to our new reality. As a designer and developer, I recognize that the interfaces we craft are more than access points to data; they are the spaces where lives unfold, where thoughts take shape, and connections are woven. In this digital ecosystem, artificial intelligence emerges as the latest bud, its roots entangled with the tendrils of our own neural pathways. AI is not just another tool; it's the mirror reflecting our quest to understand ourselves. It prompts a philosophical introspection into the nature of consciousness, a question that has been whispered through the ages from Aristotle to Descartes, and now posed anew to the coders and creators of today. These are not just systems and algorithms; they are the emergent properties of human ingenuity, a testament to our insatiable curiosity. As technology evolves, so do we, not only in our capabilities but in our identities. The tools we create are the lenses through which we view our world and ultimately, ourselves. It's in the seamless dance of form and function where the beauty of our craft lies. The elegance of code that powers a gracefully designed interface can be as profound as the notes that compose a symphony. Each line, each command, each user journey is a note played in the grand concert of human achievement. This is the essence of our time: the reciprocal evolution of humanity and technology, a duet sung in the key of innovation. As we stand at the vanguard of this ever-advancing domain, we are not just witnesses to history; we are its narrators, our creations the legacy we leave in the annals of human progress. --- ### Bridging the Gap: Integrating Empathy into UX Design https://leantonio.me/thinking/bridging-the-gap-integrating-empathy-into-ux-design · 2024-08-01 Empathy is often viewed through a purely emotional lens, a trait more suited to personal relationships than to the technical processes of UX design. However, in the landscape of modern technology, where user engagement and satisfaction are paramount, empathy emerges not only as a beneficial trait but as a fundamental component of successful design. ##### Understanding Empathy in UX Empathy in UX design means understanding the feelings, thoughts, and experiences of the user, without necessarily having experienced them directly oneself. It's about stepping into the user's shoes, regardless of one's own background or experiences, to create a product that resonates on a universal level. This empathetic approach helps designers create more intuitive and accessible products. ##### Techniques for Empathetic Design 1. User Personas: Begin with crafting detailed user personas. These personas should go beyond demographic information to include the users' emotional states, preferences, and environments. This depth helps designers foresee and address a range of user needs and scenarios. 2. Empathy Mapping: Use empathy maps to lay out what users say, think, do, and feel. This exercise pushes the design team to consider the emotional journey of users, highlighting pain points and opportunities to enhance pleasure. 3. Inclusive Feedback Loops: Establish continuous feedback loops that engage a diverse group of users. This inclusivity ensures that the product appeals to a broad audience and that minority groups' needs are considered and addressed. ##### Case Studies: Empathy in Action One illustrative case study involves a major e-commerce platform that redesigned its checkout process. Originally, the process was efficient for tech-savvy users but alienating for older or less tech-inclined individuals. By employing empathetic design principles, including user testing groups from diverse backgrounds, the company simplified its navigation and included clearer instructions, significantly reducing cart abandonment rates among all user groups. Another example can be found in a health app designed for patients managing chronic illnesses. The designers conducted extensive interviews with patients to understand their daily challenges and emotional burdens. The insights gained led to a design that was not only functional but also comforting and easy to use in daily life, thereby enhancing user retention and satisfaction. ##### Empathy Leads to Innovation Integrating empathy into UX design does more than solve user problems, it also drives innovation. By understanding the core needs and emotions of users, designers can push beyond conventional solutions to discover groundbreaking ideas that redefine user experiences. Moreover, empathetic design builds trust and loyalty, as users feel understood and valued by the brands they engage with. ##### Conclusion Empathy is more than a soft skill, it's a strategic tool in UX design that unlocks deeper understanding and connection with users. As we look to the future, the integration of empathy into design processes will not only enhance the user experience but also promote a more inclusive and accessible digital world. By embracing empathy, designers commit to creating products that are not only effective but also universally resonant and profoundly impactful. --- ### Designing UX for an AI-Driven Future https://leantonio.me/thinking/designing-UX-for-an-AI-driven-future · 2024-07-20 For decades, user interfaces have been dominated by graphical user interfaces (GUIs) built around visual metaphors like windows, icons, and pointers. User experience (UX) design has focused on making these visual interactions intuitive and frictionless. However, a monumental shift is underway as large language models (LLMs) like ChatGPT redefine how humans interact with software. Powered by cutting-edge AI, LLMs can understand and generate human-like responses, turning the software interface into an intelligent conversational agent. This disrupts traditional GUI paradigms, demanding a transformative approach to UX centred around linguistic interactions between human and AI. In an AI-driven world, the AI model itself becomes the user interface - an intelligent being that users converse and collaborate with using natural language. The fundamental principles of UX change when the interface is an active participant that can ask clarifying questions, rephrase instructions, and even challenge user assumptions. The transition from graphical user interfaces (GUIs) to conversational user interfaces (CUIs) is akin to moving from controlling software functions to developing a relationship with an intelligent assistant. User interaction patterns shift from pointing, clicking, and form-filling to an open-ended conversational exchange. Human-AI communication breaks traditional human-computer interaction (HCI) conventions. Context, tone, personality quirks, and the back-and-forth nature of dialogue become vital factors in delivering a quality user experience. Voice and text-based modalities replace visuals as the primary means of interfacing with software. UX designers must craft experiences optimised for these modalities, steeped in principles of conversation design, natural language processing, and linguistics. Challenges include handling ambiguity, asking clarifying questions, providing feedback loops, and more. New interaction flows and conceptual models tailored for CUIs will need to be prototyped and tested rigorously. AI interfaces inherently personalise interactions based on continually accumulated user data like interests, communication styles, and even emotional states. As the AI deepens its understanding of each individual through prolonged interaction, it can custom-tailor language, advice, and persona. This introduces significant ethical concerns around the autonomy, privacy, and potential manipulation of users through AI-driven personalisation. UX designers will need robust frameworks to ensure AI interfaces remain unbiased, transparent, and aligned with user wellbeing. In this radical new landscape, UX designers must move beyond crafting static visual flows and start designing for emergent conversational flows shaped by an omnipresent AI agent. Their roles and skills must evolve dramatically: - Conversation design and dialogue scripting become essential, with a focus on creating cohesive personalities and user experiences. - Interdisciplinary skills in linguistics, psychology, privacy, and ethics are vital to develop AI that influences users appropriately. - New UX methodologies and conceptual models involving simulations and reinforcement learning need to be created for conversational AI prototyping. - Collaboration between UX designers, AI researchers, linguists, creative writers, and other disciplines is key to powerful AI interfaces. To stay ahead, UX designers should immediately begin: 1. Learning foundations like conversational design, NLP, AI ethics, and related fields to build necessary skillsets. 2. Experimenting with existing conversational AI tools and open-source models. 3. Proposing conceptual AI prototypes and creative ideation exercises in their organisations. 4. Developing frameworks and best practices for emotional AI, privacy-centric CUIs, and unbiased language models. Forward-thinking companies will implement AI-driven conversational interfaces through gradual real-world pilots, data analysis of human-AI interactions, and collaborations across disciplines. Both designers and companies should explore edge cases where AI CUIs could provide significant UX enhancements. Large language models and conversational AI are rapidly reshaping human-computer interaction as we know it. Static graphical interfaces will increasingly be complemented and displaced by intelligent AI agents which users communicate with naturally. This seismic transition to AI-driven user experiences demands an entirely new set of skills and philosophies for UX designers to create cohesive, ethical, and engaging conversational user interfaces. Both UX designers and companies need to get ahead of this disruptive wave through education, experimentation, collaboration, and specialisation around AI interfaces. Those who stay rooted in traditional GUI paradigms risk getting left behind as conversational AI reshapes digital experiences and user expectations. UX trailblazers and thought leaders must embrace this incredible opportunity to innovate and reimagine user experiences in partnership with AI. The AI-driven future is approaching rapidly - the time to design for it is now. --- ### Digital Transformation: Embracing the Technological Revolution in Business https://leantonio.me/thinking/embracing-the-technological-revolution-in-business · 2024-06-15 In the rapidly evolving landscape of the 21st century, digital transformation has emerged as a pivotal force reshaping industries across the globe. No longer just a buzzword, it is a strategic imperative for businesses that aspire to stay competitive and relevant. This transformative process goes beyond merely adopting new technologies; it involves integrating digital solutions deeply into every aspect of a company's operations, customer engagement, and even its culture. By exploring some key aspects, innovation, customer experience, operational agility, and workforce empowerment, this article delves into how digital transformation is reshaping the business world. **1. Innovation at the Core** Digital transformation starts with a mindset that fosters innovation. Companies that embrace this philosophy understand the importance of continuous exploration and experimentation with emerging technologies like artificial intelligence (AI), blockchain, Internet of Things (IoT), and cloud computing. For instance, AI is revolutionising product development, enabling predictive analytics, and enhancing supply chain management. Blockchain is transforming trust mechanisms and secure transactions. These innovations not only create new products and services but also disrupt traditional business models, opening up entirely new revenue streams. **2. Redefining Customer Experience** At the heart of digital transformation lies the quest to deliver unparalleled customer experiences. Today's customers expect seamless, personalised interactions across multiple channels. Digital tools, such as chatbots, augmented reality (AR), and advanced data analytics, allow businesses to understand customer preferences better and tailor offerings accordingly. By leveraging these technologies, companies can provide real-time support, personalised recommendations, and immersive shopping experiences, thus deepening customer loyalty and driving growth. **3. Operational Agility for Competitive Advantage** Efficiency and adaptability are hallmarks of successful digital transformation. By automating processes, integrating data systems, and leveraging cloud-based solutions, businesses can significantly enhance their operational agility. Cloud computing, for example, allows for scalable resources, reducing costs and increasing responsiveness to market changes. IoT and data analytics enable real-time monitoring and predictive maintenance, minimising downtime and optimising resource utilisation. This agility is crucial in navigating today's fast-paced markets, where flexibility is key to staying ahead of the competition. **4. Empowering the Workforce through Digital Literacy** A digitally transformed organisation values its workforce's digital literacy as much as technological advancements. The integration of digital tools necessitates reskilling and upskilling employees, fostering a culture of continuous learning. This empowers workers with the skills to leverage technology for innovation, problem-solving, and collaboration. Tools like collaborative platforms, virtual training, and AI-assisted productivity enhancements not only increase efficiency but also enhance job satisfaction and employee engagement. **5. Security and Ethics in the Digital Age** As businesses digitise, cybersecurity becomes paramount. With increased reliance on digital infrastructure comes the risk of cyberattacks. A robust digital transformation strategy must incorporate advanced security measures, including encryption, multi-factor authentication, and regular security audits. Additionally, ethical considerations around data privacy, algorithmic bias, and responsible AI use must be integrated into decision-making processes, ensuring trust and compliance with evolving regulations. **Conclusion** Digital transformation is not a one-size-fits-all journey; it's a customised roadmap that each organisation must chart based on its unique needs and objectives. It's about harnessing the power of technology to drive innovation, enhance customer experiences, streamline operations, empower workforces, and safeguard against digital risks. As we continue to navigate the digital era, those who succeed in this transformation will not only survive but thrive, carving out new frontiers in their respective industries and setting the benchmark for the future of business. The ultimate goal is not just to adapt to change but to lead it, creating a sustainable competitive edge in an ever-evolving digital landscape.