Pendo for Agents brings together two sets of capabilities that help you make your AI agents more effective and understand how they perform: Agent Toolkit and Agent Analytics.
Together, they give any AI agent access to product context from Pendo—whether you build the agent in-house or use a third-party agent such as Intercom Fin. They also give your team a way to measure how the agent is used and whether it's working.
Note: Agent Analytics is generally available. Agent Toolkit is available in open beta.
What Pendo for Agents includes
Pendo for Agents has two parts. You can use either on its own, or both together.
Agent Toolkit
Agent Toolkit is the part that lets your agent act: it connects Pendo to your agent so it can use a visitor's product context in the moment. It equips your agent with capabilities such as:
- Guides. The agent surfaces a relevant Pendo guide in response to a visitor's question.
- Session context. The agent draws on a visitor's recent activity in your product when it responds.
- Feedback submission. The agent sends visitor feedback to Pendo from within the conversation.
- Workflow nudges. The agent re-engages a visitor who drops off partway through a defined workflow.
- Frustration signals. The agent responds when Pendo detects friction, such as rage clicks, error clicks, or U-turns.
Agent Toolkit connects to your agent through the Pendo MCP server and through proactive support (webhooks). For more information, see Overview of Agent Toolkit.
Agent Analytics
AgAgent Analytics is the part that lets you see how your agents are doing: it measures how visitors interact with your agents so you can find what's working and where they're not working, including:
- Usage and engagement. Prompt volume, unique visitors and accounts, retention, and reactions over time.
- Use cases. Patterns in what visitors ask, surfaced automatically or tracked as topics you define.
- Issues. Problems detected from conversation data and grouped by theme, plus issues you track manually.
- Conversations and traces. Full transcripts, tool and model detail, and session replay of a conversation.
- Experiments. Comparisons across agent versions, models, or configurations.
For more information, see Overview of Agent Analytics.
How the two parts work together
Agent Toolkit gives your agent context to act on, and Agent Analytics measures the result. You can use them independently: connect an agent in Agent Toolkit without Agent Analytics, or analyze an agent in Agent Analytics without Agent Toolkit. Using both gives you a single view of what your agent does and how it performs.
Note: Agent Toolkit and Agent Analytics are managed in separate areas of Pendo. Creating an agent in one doesn't create it in the other.
Use cases
- Give a support agent (such as Intercom Fin) the context to resolve more questions without escalating, then measure its resolution and where it struggles.
- Give an in-product assistant the ability to surface guides and re-engage visitors who drop off, then track which use cases it handles.
- Connect a custom agent you build and host yourself to Pendo, and analyze its conversations alongside the rest of your product data.