What it includes
- Task-specific structure, detail and explanation depth
- Autonomy and guidance level
- Verification behaviour and human approval thresholds
- Uncertainty presentation
- Versioned provenance on every adaptation
AI Adaptation
How should AI change for this person, task and risk?
The same person gets different AI behaviour depending on the task and its stakes. The engine resolves person × task × context × risk × organization policy and emits provider-neutral instructions.
Included in the Founding PilotWhere it stops
Model-independent. Provider connectors are adapter-ready; no native production integration is claimed.
What actually changes
Same person, low-risk drafting versus a consequential contract review
Generic
One assistant behaviour for everyone and every task: a long, evenly detailed answer, same confidence, same autonomy, regardless of who is asking or what is at stake.
NeuroBusiness
Low-risk drafting: concise, decision-first, high autonomy, single pass. Consequential review: assumptions surfaced first, uncertainty stated explicitly, checkpoint verification, and human approval required before anything is sent.
- Detail depth moves with the task, not the person alone
- Autonomy drops as stakes rise
- Verification becomes a checkpoint instead of a single pass
- Uncertainty is stated rather than smoothed over
- Human approval becomes mandatory