Verification register AI & Agents
Current reading
The readiness gap, in one scan
AI-assisted assembly · derived results
- Claimed
- 80
- Reported
- 73
- Verified
- 64
- Gap
- +16
Public ambition and stated capability
Observed practitioner reporting
Independently supported evidence
Claimed minus verified
Evidence strength Strong
Decision
What the current evidence supports
Human editorial judgment · 2026-06-28
Adopt with guardrails
- Why
- Best-in-class measured multimodal scores (81% MMMU-Pro, 87.6% Video-MMMU, 72.7% ScreenSpot-Pro, 0.115 OmniDocBench edit distance) make it the strongest VLM available, but preview status, TTFT latency, observed audio-timestamp failures, and a medical guardrail mean it needs flags, fallbacks, and domain limits.
- Next
- Pilot gemini-3-pro-preview for document/screen-understanding tasks behind a feature flag with media_resolution tuning; benchmark counting/OCR/audio-timestamping on pv's own corpus before any user-facing reliance, and keep a Flash/async fallback for latency-sensitive paths.
Constraints
Blockers
No named blocker is present in the current public projection.
Evidence summary
Derived counts
AI-assisted assembly
- Total
- 8
- Tier 1
- 0
- Tier 2
- 5
- Tier 3
- 3
- Supports
- 4
- Contradicts
- 3
- Context
- 1
- Latest observed
- 2025-12-05
Counts and dates only. Raw signals, private excerpts, trust records, and internal corpus material are not published here.
Publication record
Revisions
Initial public reading
This is the initial public reading. No earlier readiness change is recorded.