Configuring Agents

Turn agents on or off for your Org, adjust how they behave, and retire or restore them without affecting the rest of your workforce.

Your Org's AI workforce is made up of agents, and you control which of them are actually working for your Org and how each one behaves. You can switch an agent on or off for your Org specifically, add a new agent, tune settings like its icon, color, loading message, and preferred model, and retire an agent you no longer need. This page walks through each of those changes.

When to use this

Use these controls when you want your Org to stop using an agent that's otherwise available on the platform, when you want to bring an agent back that you previously turned off, when you're adding a custom agent for your Org, or when you need to change details like an agent's model preference, color, or active status. If you're assembling your team for the first time, see Building Your Agent Team for how agents are staffed and assigned roles; this page covers changing them once they exist.

Before you start

  • You need admin access to your Org to enable, disable, or edit agents.

  • Know the agent you want to change; you'll need its identifier or name to locate it in the roster.

  • For a brand-new custom agent, have its name, description, and intended purpose ready.

Reviewing your agent roster

Open your Org's agent list to see every agent available to it. For each agent you can see its name, description, color, and three statuses: whether it is enabled platform-wide by default, whether your Org has set its own override, and the effective status, which is what your team actually experiences. If your Org has not set an override, the effective status simply follows the platform default.

Enabling or disabling an agent for your Org

  1. Find the agent you want to change in your Org's roster.

  2. Set it to enabled or disabled for your Org. This creates an Org-level override that takes priority over the platform default.

  3. If you disable an agent, everyone in your Org is blocked from using it, no matter what a person's own preference is set to. An Org-level disable always overrides an individual's setting.

  4. If you enable an agent that was off by default, your whole Org can now use it, as long as no other restriction applies.

Reverting to the default setting

If you no longer want your Org to have its own setting for an agent, remove the override. Your Org then goes back to whatever the platform default is for that agent, and will automatically follow any future change to that default.

Adjusting how an agent behaves

For agents built into the platform, you can adjust whether the agent is active, whether it's visible to others, its icon and color, its loading message, and which model it prefers to use. The core description and instructions of a built-in agent are managed centrally and can't be changed through your Org's settings.

For a custom agent you add for your Org, you have full control: its name, description, instructions, the tools it can use, its model preference, icon, and color.

Adding a new agent

To add a custom agent to your Org, give it a unique identifier, a name, a description, and its operating instructions, then choose the tools it can use, its model preference, and an icon and color. The new agent becomes active for your Org as soon as it's created.

Retiring or restoring an agent

When you no longer want to keep an agent around, you can deactivate it. This keeps its history intact but stops it from being used. If an agent has never been used in a conversation, you can remove it permanently instead; an agent that already has usage history can't be permanently removed, only deactivated. A deactivated agent can be reactivated at any time, which restores it exactly as it was configured.

Example

Suppose a Support agent is enabled by default across the platform, but your Org runs its own branded support flow and doesn't want that agent available. You open your Org's agent roster, find the Support agent, and set it to disabled. From that point on, everyone in your Org sees it as off, even though the agent stays enabled for every other Org on the platform. Later, if you decide to bring it back, you either re-enable it directly or remove your override so it falls back to whatever the platform default is at that time.

After you change an agent's status, the effect is immediate: your Org's effective status for that agent updates right away, and anyone on your team trying to use a disabled agent is blocked.

Common mistakes

  • Trying to edit the description or instructions of a built-in agent. Those fields are locked; only status, visibility, icon, color, loading message, and model preference can be changed for built-in agents.

  • Expecting a permanent removal to work on an agent that has already been used. It will be refused; deactivate the agent instead.

  • Assuming a team member can turn a disabled agent back on for themselves. An Org-level disable always overrides a personal setting.

  • Leaving a stale Org-level override in place after you actually want to follow the platform default again. Remove the override rather than manually matching it.

If you're building out your Org's workforce rather than adjusting an existing agent, Building Your Agent Team covers how agents are staffed and assigned to roles. Since the agents you run affect your Org's usage, Organization Billing shows how that usage is billed and broken down.

Last updated Jul 31, 2026