Verify agent work in the source system

A poster used Hermes agents for a small product and counted tasks only after checking the real system that held the result.

Done with Hermes Agent· Intermediate· Nov 2026

What they did
They split work across several Hermes agents, each with a narrow job, and tracked everything on a board with one card per task, one owner, and one place to store proof. They would not accept a recap from the agent alone. Instead, they required evidence such as a file, a count, a live page, or a direct check in the source system, like the sent email folder or the CRM record.
How it went
That setup improved day-to-day control: fewer tasks were duplicated, less time was spent rechecking stale numbers, and fewer items were marked done too early. The source-system check also caught cases where something had been removed or never appeared. No broader performance totals or business metrics were reported.
Tips
Treat the agent’s own status as provisional until the record in the system of record matches it. For anything that could change data or contact someone, stop and ask for approval before the final action.

Try it yourself with Hermes Agent

Act as my operations agent for [small product or business]. Handle one task at a time from my board, and for each one, do the work, then verify the result in the system of record itself, such as [sent email folder], [CRM name], [platform post record], or [deployed URL]. If the task would send, submit, delete, or save anything, stop first and show me the exact action you plan to take, then wait for my approval. Afterward, give me a short status for each task: verified or not verified, plus the exact evidence you found in the source system.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on reddit.com ↗

1 more report of this

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.