Claims
Claim-level verification
Every factual claim isolated and checked against a primary source.
Hallucination & fact-check audits
Find the confident errors before your users believe them.
Claim-by-claim audits of model output by vetted experts, checked against primary sources. Every unsupported or wrong statement is logged, sourced and graded.
72-hour paid pilot · Evidence for every expert · Free replacement
What you receive
- Claim-level verification
- Citation audits
- RAG & grounding checks
- Confidence vs. accuracy
Delivered as
- Claim-level log
- Source links
- Severity grades
- Accuracy report
The problem
The dangerous errors look exactly like the right ones.
Hallucinations rarely announce themselves. A wrong citation, an invented case or a misread figure reads as fluently as the truth, and neither the model nor an automated checker reliably flags it.
In our published Urdu study, a frontier model made a mistake in about 70% of lines, and many carried no warning at all. Real accuracy was about 90%, against the 98% the model claimed.
- Fabricated citations and sources
- Confident wrong numbers, names and dates
- Self-reported accuracy that doesn't hold up
What we deliver
What you receive.
Subject experts and native speakers for the field and language of the content being checked. Delivered in your format, with a provenance record for every item.
Citations
Citation audits
Does each cited source exist, and does it say what the model says it does?
Grounding
RAG & grounding checks
Is each answer supported by the retrieved documents, or invented?
Calibration
Confidence vs. accuracy
Measured accuracy set against the model's own stated confidence.
Evidence
Error logs with sources
Each error with the source, the correction and a severity grade.
Start free
A free sample audit
Send 20 model outputs and our vetted experts audit them free within 72 hours.
Typical outputs
- Claim-level log
- Source links
- Severity grades
- Accuracy report
How vetting works
Every expert comes with the evidence.
Rather than a profile and a score, you get the record behind each person on your project, and you can review it before they start.
- Recorded AI video interview with the full transcript and a scored rubric
- ID and liveness check, with duplicate-account detection
- Credentials checked with the issuing body where the role needs them
- Integrity flags from the interview, shown to you in plain words
- Paid calibration against gold answers before any client work
How an engagement runs
From first call to scale, pilot first.
Scope
Tell us the task, the domain, the languages and your quality bar. We write the rubric with you.
NDA on requestMatch
We staff vetted experts whose field and language fit the task, and share their evidence packs.
Evidence per expertPilot
A 72-hour paid pilot on your real task, with gold items, dual review and agreement scores.
72 hours, paidScale
Scale on the model that suits you: hourly, per task or a managed pod. Free replacement throughout.
Free replacement
Related solutions
Often paired with this.
- Evaluate LLM evaluation Held-out test sets, rubric scoring and side-by-side comparisons by experts in your field.
- Experts Domain expert evaluation Physicians, lawyers, finance professionals and STEM PhDs, credentials checked at the source.
- Languages Multilingual & low-resource language evaluation Arabic (MSA and dialects), Urdu, Spanish, Portuguese and more, by native speakers.
Get started
Tell us the gap. We'll scope the pilot.
Every engagement starts with a 72-hour paid pilot. We reply to every request within two business days.
Talk it through
A 30-minute call with the founding team. Bring the task, the domain and your quality bar.
Get a quote
Share the scope in two minutes: service, team size, languages, start date and budget band.