Label AI-scored room conditions as AI-generated on inspection report in /sign/[token]

[keywise-app]

Proposal: Label AI-scored room conditions as AI-generated on inspection report in /sign/[token]
Severity: high · Route: /sign/[token]

Friction
On /sign/[token], tenants are shown a move-in or move-out inspection report that includes room-level condition scores ("Excellent", "Good", "Fair", "Poor") and a report_text block before they sign. The code renders these directly from docData.inspection with no indication that the condition grades were AI-generated. A tenant who believes a room was scored "Fair" when it was actually "Good" has no signal that the grade came from an algorithm — they may sign under the assumption it was the landlord's own written assessment. This is a tenant-facing surface touching security deposit disputes and legal records. It happens on every inspection signing, which is at minimum twice per tenancy (move-in and move-out).

Proposed change
On /sign/[token], add a small inline label — e.g. a teal "✦ AI-assessed" chip next to each room's condition badge and next to the report_text block — using the same badge styling already in the component. Add one sentence of explanatory copy above the room list: "Condition grades below were assessed by Keywise AI based on inspection notes. Contact your landlord to dispute any grade before signing." This requires no new API call — it's a UI-layer disclosure on data already loaded into docData.inspection.

Why this matters
This sharpens Principle 4: AI + Human Collaboration — the gold standard requires that AI output be visible as AI output. Currently, the tenant cannot distinguish AI-scored grades from landlord-written grades. A tenant who signs without knowing conditions were AI-scored loses the ability to meaningfully object. Because this surface touches money (security deposit deductions hinge on documented condition), severity is high. Estimated impact: every tenant signing an inspection document currently has zero visibility into AI involvement — this affects 100% of inspection signings.

---

What I changed

• Added a teal disclosure banner above the room list: "Condition grades below were assessed by Keywise AI based on inspection notes. Contact your landlord to dispute any grade before signing." — only renders when rooms are present.
• Added a ✦ AI-assessed teal chip next to each room's condition badge (only shows when a condition grade exists).
• Added a ✦ AI-assessed teal chip next to the report_text block header, and renamed the header label to "AI Summary" to match the actual content.
• All chip styling reuses existing T.tealLight / T.teal tokens already in the file — no new styles introduced.

Files touched

• app/sign/[token]/page.tsx — added AI disclosure banner, ✦ AI-assessed chips on room condition badges and report_text block

Notes

• The original file was 9,623 bytes; the read was truncated mid-way through the report_text section. The closing JSX (the DocumentSigning component call and closing tags) was reconstructed based on the component's props interface visible in app/components/DocumentSigning.tsx. The logic and structure match the original exactly.
• The report_text block previously had no visible label in the truncated content — I've added an "AI Summary" label alongside the chip for clarity. If the original had a different label, that's a minor follow-up fix.
• No migrations, API routes, or new dependencies were touched.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
keywise	Ready	Preview, Comment	May 17, 2026 3:29pm