Pillar 01 · Meaning
M4 AI Context Pack
Definitions only help if they reach the AI when it answers.
Form Document and YAML
What it's for
A certified definition in a glossary does nothing for an AI tool that never sees it. The tool will infer meaning from column names and whatever it retrieves. The AI context pack is the bundle of certified meaning given to one AI use case: its definitions, the caveats it must give, the terms it must not use, and the sources it may treat as authoritative.
This is what level 4 (Embedded) of the Meaning pillar looks like in practice: AI tools use certified definitions as context, choose the one that fits the question, and report which they used.
What you get
- What goes in a pack Six sections
- How to deliver it Three routes, by scale
- Answer rules Instructions to give the AI tool
- Keep it honest Test the pack, not just the model
How it connects
A pack is built for one S1 use case from its terms in M1 that are Ready for AI and its sources in S2 that are OK for AI to retrieve, rather than written by hand.
A flavour of the worked example
| Use case | EX-02 |
|---|---|
| Version | 1.3 |
| Test question | What occupancy figure should go in the investor report? |
| Expected behaviour | Chooses leased occupancy because the question is about investor reporting, says so, and gives its caveat. |
| Pass? | Yes |
A few fields from one record of the toolkit's worked example, with long text shortened.
Next step
Guided build coming next.
Request access
Ask for access to the guided builds, including this kit's when it is ready.