Verification register AI & Agents
Current reading
The readiness gap, in one scan
AI-assisted assembly · derived results
- Claimed
- 90
- Reported
- 90
- Verified
- 83
- Gap
- +7
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
- Apache-2.0, a full size ladder, a mature ecosystem (4.2M Ollama downloads, OpenRouter availability, a 64-author technical report) and strong perception/agent results — corroborated by the independent OmniDocBench leaderboard (89.78 overall) — make it the lowest-risk, most flexible VLM of the three; only caveats are benchmark transparency and extreme-context claims.
- Next
- Pilot a small/mid variant (8B or 30B-A3B) for document/chart/OCR + GUI-agent tasks via a hosted endpoint, validate on our own data given the modified-prompt caveat, then decide on self-hosting.
Constraints
Blockers
No named blocker is present in the current public projection.
Evidence summary
Derived counts
AI-assisted assembly
- Total
- 7
- Tier 1
- 0
- Tier 2
- 2
- Tier 3
- 5
- Supports
- 5
- Contradicts
- 1
- Context
- 1
- Latest observed
- 2026-03-31
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.