Agents overview
What an agent is, the pages you use to configure it, and its lifecycle from draft to live.
An agent is the thing you build in FluentChat — a configured AI bot with its own persona, knowledge, actions, channels, and settings. You can run many agents in one workspace, and each is fully isolated from the others.
The agent workspace
Open an agent and you'll find a set of pages along the side:
Playground
A live chat against your real agent, plus widget appearance controls.
Instructions
The persona and rules that shape how the agent behaves.
Data sources
The knowledge the agent answers from.
Actions
The tools the agent can call mid-conversation.
Channels
Where the agent is deployed.
Settings
Working hours, handoff, off-topic replies, images, voice, and more.
The agents list
The Agents page lists every agent in your workspace. Each card shows the agent's name, a status pill, and the packs it has installed. From the card's menu you can duplicate or delete an agent, and you can filter the list by status.
Agent status
An agent moves through a simple lifecycle:
Prop
Type
Your plan sets how many agents a workspace can have. When you reach the limit, the New agent button prompts you to upgrade. See Plan features.
Isolation
Every agent is a separate island. The knowledge you add, the actions you install, the store you bind, and the cached answers the agent produces all belong to that one agent. Ask Agent B about something only Agent A knows and it simply won't have it — by design. This keeps a Sales agent and an internal HR helper safely apart in the same workspace.