Emet launches read-only verification product for AI answers
Understand this story
Emet has released an AI answer verification platform that, after an agent's answer is issued, reruns the query in a read-only manner, re-derives every number in code, and checks the results sentence by sentence. Emet says the platform does not enter the request path, and access permissions are controlled by the user, scoped, and revocable at any time. In an official example, an AI report gave Q2 ROAS as 3.46x, and Emet recalculated it as 2.43x, because the original query covered only June 16 to 22 rather than the entire Q2.
AI overview based on source reports · Updated 10 hr ago
Coverage timeline
Follow the coverage to explore different perspectives.
- Hacker News AIEmet: Code checking AI. Not another model grading the answer
Emet is a read-only verification product that runs after an AI agent's answer is sent, re-running the query on your source data and re-deriving every number in code to produce an accuracy report and Slack alerts. It checks date ranges, attribution, time zones, currency, math and claims against live data rather than having a model grade its own output, and it runs alongside tracing tools like Langfuse, LangSmith and Arize Phoenix.
Activity for this story
- Current comparable activity
- 8
- Peak comparable activity
- 10Nov 18 00:00
- Change in the past 24 hours
- –
Trends compare the same fully observed participants, which may cover fewer participants than current activity. Move the pointer or tap the chart to inspect hourly activity; use the left and right arrow keys with a keyboard.