Glossary¶
Every term the rest of the documentation relies on, defined once. Keep this open in a second tab.
Addendum¶
A short note a surveyor can add to an auto-drafted reliability file note to clarify or qualify it. The original draft is never overwritten — the addendum sits alongside it.
Alert¶
A real-time exception flag raised when a classified observation trips a detection rule — restricted data in an unapproved place, an anomalous spike in volume, a tool used outside its purpose, an opt-out breach. Alerts go to the relevant supervisor and the AI Lead. They are awareness, not a queue.
Audit report¶
A quarterly or annual compilation — observation counts, classification mix, sign-off rates, dip-sample results, risk-register status — produced for an auditor or for the firm's own assurance.
Claude Cowork¶
Anthropic's Claude running on your firm's Microsoft 365. It's where the work happens. Chronity sits alongside it, not inside it.
Classification¶
The overnight process that assigns each observation a track and a sensitivity by matching it to your firm's signed taxonomy. See How it works.
Client opt-out¶
A client who has declined the use of AI on their matters. Admins keep a keyword list (company name, property names, email domains); Chronity scans every AI action against it, hard-blocks client-facing actions on a match, and warns on everything else.
Code Mode¶
The pattern by which Claude sees only three tools on the wire (get_context,
search, execute) instead of dozens. It keeps the connection lightweight; you
never need to think about it.
Daily Check¶
A weekday-morning probe that proves Chronity's Microsoft 365 write path still works, by creating and instantly deleting a throwaway draft. Its history shows on the Daily Health page.
Daily Health¶
The user-portal page showing the recent results of the Daily Check — a green run of dots when the write path is healthy.
Digest¶
A once-a-day email (07:00 your time) summarising the previous day's AI activity in your scope. Sent to supervisors and admins. Skipped if there's nothing to report.
Dip-sampling¶
The Standard's preferred way to assure high-volume work: instead of reviewing every Track 2 output, Chronity selects a defensible random sample each quarter for a reviewer to check. See AI Lead → Dip-sampling.
Due-diligence pack¶
A generated, downloadable document describing what Chronity processes, how, where, and under what encryption — for the firm to drop into its own supplier due-diligence file.
Heartbeat¶
An independent liveness signal: Claude pings Chronity at the start of every conversation, separately from the MCP connection. Pings arriving but observations not is the fingerprint of a dropped connection, and triggers an admin alert.
Material Impact Assessment (MIA)¶
Your firm's written assessment of the impact of each AI use case — the document the RICS Standard requires. In Chronity it lives as the taxonomy.
MCP¶
Model Context Protocol — the open standard by which Claude connects to Chronity. You only meet it as a connection URL during setup.
Observation¶
The atomic record of a single AI-assisted action — the thing the whole pipeline is built on. Its summary is encrypted with your firm's key the moment it's created.
Observation pipeline¶
The end-to-end flow from capture through classification to the audit trail. The single most important concept — see How it works.
Read-only mode¶
What happens if a subscription falls more than seven days past due: the AI connection stops, but the portal keeps working so the firm can fix billing without losing the audit record.
Reliability file note¶
The written reliability decision the Standard asks for: assumptions, concerns, mitigations, conclusion, in the first person, naming the responsible surveyor. Chronity auto-drafts one for every Track 1 observation. Also called a reliability note.
Sensitivity¶
How sensitive the content is, on a four-tier scale: public, internal, confidential, restricted. Assigned during classification from your signed taxonomy.
Sign-off¶
Under Chronity's passive model, a surveyor sending work out under their own name is the professional sign-off. The legacy sign-off queue (Track 1 items awaiting a qualified reviewer) still exists during the transition but is being retired.
Standing reliability analysis¶
The reusable block your taxonomy holds for each use case — assumptions, known reliability concerns, standard mitigations, conclusion — which the auto-drafted reliability note quotes by reference. Signed once, applied automatically.
Super-admin¶
Chronity's own operators, for provisioning and support. They never see your client content in plain form.
Taxonomy¶
Your firm's signed MIA expressed as a version-controlled list of use cases, each with a default track, default sensitivity, a standing reliability analysis, and an MIA section reference. The classification authority — see AI Lead → The MIA & taxonomy.
Tenant¶
Your firm's isolated space in Chronity — its own database, its own encryption keys, its own branding. One firm, one tenant.
Track¶
The sign-off route an observation takes, assigned from your taxonomy:
- Track 1 — formal deliverable. The firm takes professional responsibility; a named qualified surveyor stands behind it; an auto-drafted reliability note is produced.
- Track 2 — professional workflow. Routine work using judgement; assured by quarterly dip-sampling, not output by output.
- Track 3 — non-material. Platform plumbing, terminology lookups. Logged for completeness only.
Use case¶
A single entry in the taxonomy — client email drafting, rent review cross-checking, and so on — carrying its own default track, sensitivity, and standing analysis.