What It Is

The Evidence Layer Under the AI Tools

As of today, the library holds 79 risk records. Each one documents a single operational failure mode: the pathway it occurs on — workplace, public services, clinical, policing and justice, education, or residential settings — the setting, who typically raises the challenge, the mechanism that causes harm, and the mitigation that stops it. It is the fourth catalogue estate alongside Authorities, Personas and Scenarios — built to make an AI answer concrete about how things actually go wrong, not to dramatise a situation.

What It Does

Three Sample Records

Every record is written in the same short, clinical register — a pattern, not a story. Three representative samples, out of the full library:

Workplace toilet gatekeeping (colleague complaint → HR restriction)

Pathway
Workplace
Setting
Workplace — toilet
Mechanism
Misinterpreted guidance; institutional gatekeeping.
Mitigation
HR applies a behaviour-based standard — no ad-hoc restrictions; challenges are treated as harassment.

GP reception registration: "need sex at birth" and visible outing

Pathway
Clinical
Setting
Primary care — front desk
Mechanism
Records propagation; administrative gatekeeping; public outing.
Mitigation
Move the discussion private; register using correct details; collect only clinically necessary fields.

Leisure centre registration and membership data mismatch outs a participant

Pathway
Public services
Setting
Leisure centre / gym front desk — reception
Mechanism
Data mismatch; front-desk outing; service-delivery failure.
Mitigation
Move the conversation private; correct visible details where possible; avoid public questioning.

Where It's Used

Inside the Tools — and Browsable in the App

Risk records are retrieval evidence for specific AI tools — Roast My Policy, Ask Quinn, andProportionality Check draw on them when a policy or a decision needs testing against a documented pattern of how things actually go wrong, not a hypothetical. Signed in, you canbrowse the full library in the app — and you meet it inside those answers, where each record appears as a named, clickable citation.

Not the same as Scenarios

Scenarios is the narrative rehearsal library —Flashpoint walks you through one step by step, with the personas as the cast. The risk library is deliberately separate: it states the pattern, it does not tell the story, and Flashpoint does not draw on it. The two are siblings, not one product.

Who It's For

Who the Risk Library Serves

HR and EDI leads who want a policy critique grounded in a documented pattern rather than a guess, governance leads building a proportionality case who need the operational failure mode named precisely, and anyone using Ask Quinn for a question that turns on how a situation typically goes wrong. Access to the tools it grounds scales with your plan — see the full feature comparison — and you canstart free to see it working inside an answer.