Business AI Agents for recurring work

Business AI requires you to connect your own account with a supported AI provider.

Give each Agent a recognizable name and persistent instructions. Add relevant company knowledge and supported reference files. Run it manually or set a schedule for automatic execution.

An Agent can perform authorized business operations and supported Integration Builder actions. Choose tasks that fit your company’s enabled capabilities and configured connections.

Each Run has its own conversation and execution context. Watch progress through a chat-style interface and inspect the findings, actions, generated files, and history it preserves.

For example, configure an Agent to review permitted operational records and prepare a recurring summary. The scope and available follow-up actions depend on your company’s configuration and permissions.

You can observe an active Run and stop it. Additional instructions and files can be provided when a Run reaches an eligible waiting-for-review, blocked, interrupted, or failed state; you can then continue it.

Continuation is a response to a Run that needs input or recovery. It is not continuous messaging into an actively executing Run. Pause an Agent to prevent future execution, or stop an active Run when needed.

Runs execute with the company’s authorized automation access, separately from your signed-in chat session. Current permissions and authorization for each operation still apply; choosing an Agent’s controlled actions is not its entire access boundary.

Creating an Agent does not grant unrestricted access to company data. Human-only approval, rejection, and reconciliation decisions remain human-controlled.