Configure AI Agents for your Account

Overview

The Configuration page holds the settings that apply to every AI agent on your account, no matter who created it.

It has three sections:

• Global instructions — context that every agent gets on top of its own instructions, for example your company's terminology or standards.
• Notifications — whether agent owners are told when a run fails or an agent is paused automatically.
• Permissions — who on your account can build agents.

Read more about building and running agents in Create & Manage Your AI Agents.

Read more about Privacy & Security for AI Features.

How to Access AI Agents Configuration

1. Click AI Agents
2. Click Configuration
   

Set Global Instructions for All Agents

Global instructions are context you write once for the whole account. Every agent reads them in addition to its own instructions, so this is the place for company rules, terminology and standards you would otherwise repeat in every agent.

For example:

• "We are a general contracting company based in Vienna."
• "Always respond in German, using a formal tone."

You can also start from one of five ready-made sentences:

• Company context
• Language preference
• Terminology
• Primary use case
• Industry regulations

To set the global instructions:

1. Under Global instructions, write the context that applies to all your agents.
   
2. Optional: Click one of the sentences next to Start from: and complete it. Each one is added at the end of what is already written, so you can use several.
   
3. Optional: Click Rewrite with AI to get an AI suggestion with clearer phrasing.
   
4. Optional: You can keep the suggestion, Undo it, or write your own version instead.
   
5. Click Save instructions
   

Global instructions can be up to 5,000 characters, and changes apply on the next run of each agent. Leaving them empty is fine — agents then work from their own instructions only.

Global instructions are context for the agent, not a rule it enforces. They cannot limit what an agent is allowed to read, write or send — that is set per agent through its project, triggers and actions.

Rewrite with AI consumes your AI usage limit. Read more in AI Usage Limits and Upgrades.

Choose Which Agent Notifications Are Sent

Agent owners are told when something goes wrong with one of their agents. The two settings here switch those messages on or off for the whole account — they are not personal settings, so a change affects every agent owner.

Both are on by default:

• When an agent run fails — a run ended without the agent finishing its work.
• When an agent is auto-paused — the agent stopped itself and needs to be activated again.

To switch a notification off:

1. Under Notifications, click the toggle next to the event.
   

The change is saved immediately and applies to everyone on the account. Agent notifications are sent to the Message Center in the Webapp only — there is no email and no mobile notification for them. Read more in Message Center & Notifications in the Webapp.

Read more about why an agent stops itself in Create & Manage Your AI Agents > When an Agent Pauses Itself.

See Who Has Access to AI Agents

Permissions lists everyone on your account who holds the user permission AI Agents.

To search for users in the list:

1. Enter their name
2. Next to the search field you see the number of all users with the permission.
   

The permission itself is granted in user management, not here. To open a user's settings:

1. Under Permissions, click the Edit user in user management icon next to the user.
   

The icon is only shown if you have permission to manage users on the account.

Read more in Configure User Permissions.