Guide
Connecting custom agents in one AI workspace
Connecting custom agents in one AI workspace means bringing the agents and tools your team builds onto a single plane alongside built-in capabilities, so they share the same interface, context, and controls. It beats scattered point tools because everything is orchestrated together rather than living in disconnected silos. The result is one governed surface instead of a dozen unmanaged logins.
The problem with scattered agents
Most teams accumulate agents the way they accumulate browser tabs. A research bot lives in one app, a coding helper in another, and a custom workflow agent somewhere on an internal server.
Each tool has its own login, its own data trail, and its own blind spots. Nobody can see the full picture, and security can't enforce a consistent policy across all of them.
What 'one plane' means
Bringing agents onto one plane means they run inside the same workspace, share context, and can hand work to one another instead of forcing people to copy results between disconnected apps.
Your custom agents sit next to chat, deep research, web search, and image generation. They're orchestrated together, so a single request can draw on whichever capability fits the task.
Governance for agents
A shared plane is only useful if it's governed. Every agent should inherit the same access rules, the same guardrails, and the same audit trail as the rest of the workspace.
That means SSO for sign-in, RBAC for who can use what, guardrails on behaviour, and audit logs that capture activity across built-in and custom agents alike.
How ChatLite does it
ChatLite lets you connect your own custom agents and tools into the same workspace as built-in chat, deep research, web search, and image generation. They're orchestrated together rather than bolted on.
Everything is governed with SSO, RBAC, guardrails, and audit logs, and ChatLite can be self-hosted and customised to fit your environment. Learn more on the enterprise page or explore the full features overview.