Agent Toolkit is where you set up and equip an AI agent with Pendo capabilities, so it can guide, support, and act on behalf of your users.
Agent Toolkit brings your AI agent and its Pendo capabilities together in one place. Instead of configuring each capability separately, you start with an agent and attach the capabilities you want it to use. It supports two agent types: agents that run on Intercom Fin, and custom agents, which are any non-Fin agents you connect directly to Pendo, whether you build it in-house or configure a third-party agent.
You don't need Agent Analytics to use Agent Toolkit. You can create and connect agents through Agent Toolkit on its own. Creating an agent in Agent Toolkit doesn't create one in Agent Analytics, and deleting an agent from Agent Toolkit removes it only from Agent Toolkit.
Note: Agent Toolkit is available in open beta. It's part of Pendo for Agents, which brings Agent Toolkit and Agent Analytics together. For more information, see Overview of Pendo for Agents.
Use cases
Connect an agent in Agent Toolkit to:
- Surface relevant Pendo guides when a user asks your agent a "how do I…?" question.
- Give your agent session context, so it can draw on what a user has recently done in your product when it responds.
- Deliver proactive messages in response to signals such as user frustration.
- Deliver proactive messages when a network request matches a rule you define, so your agent can reach out with context and offer help.
- Deliver proactive messages when a user drops off from a specific workflow, to encourage them to complete it.
- Let visitors submit feedback to Pendo directly through the agent conversation.
How it works
An agent in Agent Toolkit connects to Pendo through two types of connection, and each one unlocks a different set of capabilities:
- MCP connection. Powers guides, session context, and feedback submission.
- Proactive support connection. Delivered through webhooks. Powers workflow nudges and frustration signals.
You complete one or both connections as part of the agent setup. The steps you see depend on your agent type. For setup instructions, see Set up an agent in Agent Toolkit (Intercom Fin) or Set up a custom agent in Agent Toolkit.
After setup, you enable and manage each capability from its tab on the agent:
- Guides: set when each guide launches, including target segment and target Page, then enable it for your agent. A guide must be eligible and meet its conditions before your agent can use it.
- Workflow nudges: define a workflow that visitors are expected to complete. If a visitor drops off partway through, the agent nudges them to finish.
- Frustration signals: toggle individual signals on or off and set the scope (Pages or Features) for each signal type.
- Session context: for Intercom Fin, enable the Pendo data connectors in Fin, then confirm and save. For custom agents, enable session context directly in Agent Toolkit.
- Feedback submission: for Intercom Fin, enable the Pendo data connectors in Fin, then confirm and save. For custom agents, enable feedback submission directly in Agent Toolkit.
On the AI agent's Overview page, capabilities are grouped under In-conversation help (session context, feedback submission, and guides) and Proactive help (workflow nudges and frustration signals). Select Set up on a capability to turn it on. To remove an AI agent, open it and select Delete AI agent. Deleting removes the AI agent from Agent Toolkit only and can't be undone. If it's also in Agent Analytics, it stays there. Its service account, webhook, and Intercom connection stay active until you turn them off in Settings > Integrations.
Next steps
- Set up an agent in Agent Toolkit. Add an Intercom Fin agent to Agent Toolkit and connect it to Pendo.
- Set up a custom agent in Agent Toolkit. Connect an AI agent you build in-house or a third-party agent you configure.