Your best people carry knowledge they can't easily write down. The Workbench sits with them and, through voice-first AI dialogue, extracts not just what they know but how their knowledge connects — turning expertise into structures that others can genuinely learn from.
Traditional knowledge capture creates manuals, wikis, and process documents. These record what an expert said but not how they think. The connections, derivations, and judgment that make their knowledge valuable are lost.
Knowing that topic A relates to topic B is taxonomy. Knowing how to get from A to B — and back again, through derivation and explanation — is understanding. Most knowledge tools give you the first. Learners need the second.
The most valuable knowledge surfaces in conversation, not documentation. An expert explaining to a curious, probing questioner will articulate things they've never written down and didn't know they knew.
An AI Informed Assistant conducts structured voice dialogues with your subject matter experts. It doesn't just record — it probes for derivation, checks for connections, insists on cyclicity, and surfaces the structure behind expertise. The result is a navigable knowledge architecture grounded in Gordon Pask's Conversation Theory.
Scope the domain, identify the experts, define who needs to learn it. One or two meetings.
4–10 voice sessions per expert, 25–30 minutes each. The expert talks. The AI captures structure in real time.
Between sessions: cyclicity testing, analogy formalisation, gap identification. The emerging mesh is refined and verified.
A complete entailment mesh — all derivations, analogies, and descriptors. Optionally: a Wayfinder learning instance for your people.
The Workbench builds entailment meshes — cyclic, richly connected knowledge structures where every idea can be reached from every other through derivation. This is what makes knowledge genuinely learnable rather than merely browsable.
Experts talk naturally. The AI does the structural work — probing for derivations, reversals, analogies, and models. No templates, no taxonomies, no filling in boxes.
The result is not a concept map of what goes with what. It is an entailment mesh — how to derive any idea from any other, in both directions, through explanation and construction.
A domain that covers everything but connects nothing is a reference shelf. Cyclicity guarantees every path leads somewhere and every idea can be reconstructed from multiple starting points.
Each client gets their own dedicated database, their own domain, their own environment. US or UK data residency. Your expert knowledge stays where you need it to be.
Whether it's clinical reasoning, engineering practice, financial regulation, or any domain where understanding matters more than recall — we'd like to hear from you.
peter.tuddenham@coexplorer.com