GRID Observatory

What is GRID?

GRID (Governance Risk and Infrastructure Diagnostics for SEND-AI Environments) is a computational governance framework for non-verbal SEND settings. It addresses a structural gap: non-verbal autistic learners cannot self-report safeguarding concerns. Their behaviour is the only disclosure mechanism. But sub-threshold behaviour rarely reaches the institutional record under narrative reporting and high workload.

What this Observatory shows

One classroom across a working week. The Focal Student is a non-verbal learner. Trained and untrained staff move around the room. The Behavioural Lead and DSL sit in their adjoining offices. Three governance conditions can be selected. Watch what reaches the record, and how the Focal Student's stress trajectory differs across architectures.

Why the building view matters

Under GRID, a flag travels from the classroom to the correct office (DSL for cumulative harm, Behavioural Lead for trigger isolation). Under Naive AI, an alert flashes the whole room. Under Baseline, nothing structured leaves at all. The architecture difference is visible.

What am I seeing?

Interact: Click any agent for an info card. Scrub the timeline. Step forward or back. Switch governance conditions. Toggle Compare to see all three side by side.