← The workWork for Tonic
Tonic website prototype
Tonic wanted its own site to demonstrate the kind of intelligent digital experience it builds for clients.
Work for Tonic itself: a proposed new agency website. Spring 2026.
His role
- StatedProposed, designed and prototyped it.
- DocumentedAll commits in the repository are his.
His contribution
- DocumentedRule-based personalisation: visitors are sorted into intent states (exploring, evaluating, ready to buy, returning) from behaviour signals, held locally.
- DocumentedAn 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.
- DocumentedAn experience map relating user groups, needs, pages, components and journey rules.
Decisions and trade-offs
- DocumentedDeterministic routing and personalisation, with the model confined to grounded answers; CRM hints off by default.
What was delivered
- DocumentedA working prototype. Whether it went live is not established.
What is documented
- UnknownNo outcome, metric or business result is documented for this work.
What this demonstrates
- InferenceWhere intelligence sits in a product, and what stays deterministic, decided explicitly.
Fullest account: No public case study yet.
Each line is labelled by the evidence behind it. Documented: checkable in code, a live product or a primary artefact. Stated: his own account. Attributed: a wider team's or client's. Inference: a reading of the evidence, not a fact. Unknown: not established. No outcome or metric is documented for any of this work. The same record is published for AI assistants at /llms-full.txt.