Your agent’s brain: knowledge, judgment and supervision
An agent answers well when it knows how to behave, what information to use and when you should review it.
An agent answers well when it masters three things: how it should behave (judgment), what information it can use (knowledge) and when you should review it (supervision).
Judgment and instructions
On the agent page, the Identity tab holds its personality and instructions. There you define its tone, which questions it asks to qualify and when it should escalate the lead to a person on your team.
The company brain
Under AI Agents > Brain you have a knowledge library shared by all your agents. Upload your listing sheets, terms, areas and FAQs: your agents use them to answer with real data, without making things up.
Supervision
At first the agent does not need to send on its own. In Conversations you review its suggested replies and approve, edit or dismiss them before they go out. As you build trust, you give it more autonomy.
Add only reliable, up-to-date information to the brain. If something changes -a price, a term-, edit it there and all your agents use it instantly.
Related articles
- Create your first AI agentGive a name, channels and personality to an agent that answers and qualifies your leads around the clock. Three steps.
- How your agent answers and qualifies leads 24/7Every lead gets an instant reply in its channel, with the right questions to gauge intent and what they want.
- The leads inbox: conversations, review and reply approvalIn Conversations you see every chat with your leads and decide what your agent sends: approve, edit or dismiss its replies.