This article covers how to create and configure Agents, manage their settings, and monitor their activity.

👉 Start with AI Agents: Overview for an introduction to what Agents are, how they differ from the AI Assistant, and how to enable them for your organization.

Agents only have access to objects that have been explicitly shared with them — projects, budgets, deals, and other records must be shared with the agent directly for it to work with them.

This applies whether the agent is running on a schedule or being used through the AI Assistant.

Starting a conversation with an agent does not grant it access to your data — it can only see and act on what has been shared with it.

📌 If an agent isn't performing as expected, check what's been shared with it. Missing access to a project or record is the most common reason an agent can't complete their task.

To create an Agent, go to Resourcing > Agents. This is where all your Agents are listed.

To add a new one, click + Agent.

Enter a name, choose an avatar, and describe what you need the Agent to do. Then click Create Agent.

The AI Assistant will generate the Agent based on your description, assign it a permission set, and set up any triggers it can. Depending on your prompt, it may ask follow-up questions to complete the setup.

📌 You can also open the AI Assistant directly and ask it to help you create an Agent.

You can view and manage all your Agents from the Agents list.

Adjust the view by adding fields and applying filters the same way you would in other list views in Productive.

Click on any Agent to open its profile. The profile is organized into four tabs: Info, Activity, Projects, and Permissions.

The Info tab contains the Agent's details, triggers, instructions, connectors, and skills.

Manager — The person responsible for the Agent, usually its creator. Used for the organizational chart and any agent-related approval flows. Only the Agent's manager can edit its settings.

Triggers define when the Agent runs automatically, without the need to start it manually.

To create a new trigger, click + Trigger.

You can set a daily, weekly, or monthly schedule and add specific instructions for each run.

To show all runs for a trigger, open the three-dot menu next to it and select Show runs. To run it manually, select Run now.

Instructions define how the Agent behaves, what rules it follows, what it focuses on, and how it handles specific situations.

The Agent's initial instructions are written by the Assistant during setup. You can edit them manually here at any time, or update them by prompting the Assistant directly.

Connectors give the agent access to external tools and services. Once connected, the agent can use them to read data or take actions in those systems as part of its work.

To add a connector, go to the Info tab and click + Add connector. You can disconnect a connector at any time from the same section.

📌 You can only add connectors you've already connected to your own account.

📌 Connectors are added and managed by the agent's manager only. Users the agent has been shared with can see which connectors are active but cannot add or remove them.

👉 To connect your account to a connector and learn how to manage them, see Connectors.

Skills are reusable instruction sets that extend how an agent behaves — additional context, process guidelines, or domain-specific rules that don't need to be part of the main instructions.

To assign a skill to an agent, click + Skill on the Info tab and choose from the list. An agent can have multiple skills at once, and all active skills are always on — the agent follows them on every run, whether triggered on a schedule or started through the AI Assistant.

👉 For more on creating, managing, and sharing skills, see What Are Skills?

The Activity tab shows what the Agent has been doing. It's split into three sections:

Use this to stay informed about what the Agent is doing and review its activity history at any time.

Shows the projects the Agent has been added to as a member. You can also add the Agent to additional projects directly from this tab.

The Permissions tab is where you assign permissions to the agent. This determines what types of actions the agent is allowed to perform — creating tasks, updating records, managing time entries, and so on.

To configure the agent's permissions, click Edit Permissions.

The available permission sets are limited to those the agent's manager holds. If the manager doesn't have a certain permission, it won't be available to assign to the agent.

📌 If a permission is assigned but the manager later loses it, that permission will no longer apply. You'll see a warning directly on the affected permission.

👉 For a full list of available permissions, see Permission Groups and Permissions.

By default, Agents are private — only you can see and use them. To make an Agent available to others, click Share on the Agent's profile.

Select the users you want to share the agent with and click Add.

Click the three-dot menu on the Agent's profile to access additional actions:

From here, you can: