Training · Knowledge systems

Course material you can question.

Knowledge systems built from three professional curricula: an applied-AI consulting curriculum, an AI project-management curriculum (CPMAI®-aligned), and a project management diploma program. Each answers the way the material would, and points back to its source.

Shipped to a client
DomainProfessional training
Corpora3 curricula
StackClaude · pgvector · Supabase
AnswersGrounded, with sources
6,912Passages from 196 lessons in the applied-AI curriculum
1,918Passages from 105 documents in the PM knowledge system
3Curricula, one method
SourcesEvery answer points back to the material
01
The problem

The knowledge survived the course. The access didn't.

Professional programs produce hundreds of pages of frameworks, gates and practice guidance, then hand them to people whose real questions arrive months later, mid-project. When finding the guidance costs more than guessing, people guess.

Keyword search does not fix it. "Should this project use AI at all?" is answered by a decision framework spread across a module, not by the page that contains those words.

Pain pointWhat it costWhat the build does
Guidance trapped in binders
Retrieving it costs more than guessing.
Each curriculum turned into a system you ask in one question.
Search finds words, not reasoning
Fragments of frameworks whose logic spans a module.
The material's structure (criteria, gates, exceptions, order) made explicit.
Drift from the source
Generic assistants blend the course with the open web.
Answers grounded in the material, with sources.
02
The approach

Structure first. Retrieval second.

Each curriculum is broken into its real units (concepts, rules, decision points, gates) and mapped to the situations each one governs, before anything touches a search layer.

A knowledge base is not a pile of documents with search on top. It is the material's decision structure, made answerable.

With the structure explicit, a question routes to the governing framework, the answer follows that framework's reasoning, and the response points back to the source so faithfulness can be checked. Content is tagged by depth: the facts, how things are done, and the judgment that comes from practice.

Three curricula with nothing in common took the identical build. That is the point: it is the same pattern behind every "our expertise lives in binders" engagement.

03
How it works

Four stages from corpus to consultable system.

The same sequence for training material today and a firm's internal expertise tomorrow. Click a stage.

04
What it shows

What this work shows.

Expertise stops walking out the door when it stops living in documents.

Structure beats search

Faithful answers come from encoding the material's decision structure, not from better search over raw text.

One method, any domain

Three unrelated curricula, one build.

Design for the moment of need

Built backward from the mid-project question, not forward from the table of contents.

Proven, then generalised

The same pattern now powers the Company Brain for firms.

05
See it working

Try something like it.

ClaudeOpenAI embeddingsSupabase Postgres + pgvectorDepth tags: facts, how-to, judgment

What does your organisation know that nobody can find?

The same build fits internal standards, methodology binders and onboarding programs. Start with the question your team asks most and answers worst.

📅 Book a call