# How do you make sure the agent did its job in the downstream? — retained record

> Retained reddit record from reddit:AI_Agents.

Canonical: https://highsignal.app/data/records/d1841ebf26f43775

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:AI_Agents)  published 2026-11-17 12:06 ingested 2026-11-18 02:36

[original source ↗](https://www.reddit.com/r/AI_Agents/comments/1xyirql/how_do_you_make_sure_the_agent_did_its_job_in_the/)

## retained text



```
we run agents that take real actions: booking appointments, issuing refunds, updating CRM records. We use Langfuse for tracing so we can see every LLM call, tool call and tool response.

What we cannot see is whether the action actually landed correctly in the business system, for example, the API returned 200, but the record is not in the crm.

from what I can tell, Langfuse code evaluators run without network access, so they can only judge what’s in the trace. Braintrust scorers look like they might allow outbound calls, but I haven’t seen anyone use them to check against a CRM or Stripe.

so a genuine questions for people running agents in production:

1. Is there a tool that does this, i.e. checks the trace against the system of record, not just the trace content?
2. If you built something yourself, what does it look like? A reconciliation job, a scorer, something else?
3. Or is this just not a problem worth solving in your setup?

Would love to hear how others handle it.
```



document fetched 2026-11-18 00:19 · canonical https://reddit.com/r/AI_Agents/comments/1xyirql/how_do_you_make_sure_the_agent_did_its_job_in_the
