How it works
Anywhere you're chatting with an agent, you can tag it and give direct feedback in plain language. For example:
@goava this company is not relevant for me since this industry doesn't sell outbound.
That's it. No settings menu, no separate feedback form. The moment you send that message, the agent registers the correction and stores it in its memory.
What happens to your feedback
The feedback doesn't just disappear into a log. It gets added to the agent's memory, which means it actively shapes how the agent behaves going forward. The next time it's building a prospect list or evaluating companies for you, it will factor in what you told it — in this example, steering away from industries that don't fit an outbound sales motion.
Over time, this turns into a compounding effect: the more feedback an agent receives, the sharper and more tailored its output becomes to how you and your team actually work.
Why this matters
Agents don't have perfect context on your business from day one. They learn your preferences, exclusions, and priorities the same way a new team member would — through ongoing correction and guidance. Giving feedback in the chat is the quickest way to close that gap, and it takes seconds.
A few tips for good feedback
A little specificity goes a long way. Instead of just saying something is wrong, explain why — that reason is what the agent actually learns from.
- "This company is not relevant since this industry doesn't sell outbound" teaches the agent a rule, not just a one-off exception.
- "Skip companies under 10 employees, we need more scale" teaches a filter it can reuse.
- "This contact already responded no last quarter" teaches it to recognize a pattern, not just avoid a name.
The clearer the reasoning, the more useful the memory becomes — and the less you'll need to repeat yourself next time.