PostHog scouts Bring Scheduled, Learning Agents to Product Analytics.

PostHog is adding scheduled AI agents that can act on users’ behalf, pushing its analytics platform beyond answering questions and toward carrying out recurring work. The new feature, called scouts, can run on a schedule and share knowledge through a common scratchpad, allowing agents to retain useful context over time.

Paul D’Ambra, a member of PostHog’s product team, disclosed the feature on LinkedIn. His post offered an early description rather than a full product announcement, so important details such as pricing, plan requirements and general availability have not yet been published.

Even with those gaps, scouts signal a meaningful change in how PostHog expects teams to use product data. Instead of requiring a founder, analyst or product manager to open a dashboard and investigate an issue manually, an agent could be assigned work that runs repeatedly without a fresh prompt each time.

Scheduling is the central distinction. Many analytics assistants operate reactively: a user asks a question, the system queries available data and returns an answer. PostHog scouts are described as agents that wake up on a schedule and perform actions for the user.

PostHog has not yet documented the complete set of actions scouts can take. Based on the announcement, the feature should therefore be understood as an automation layer rather than a specific collection of prebuilt reports. Potential applications could include recurring checks of product metrics, periodic investigation of unusual behaviour or routine analysis, although PostHog has not confirmed individual workflows.

For small product teams, the practical attraction is straightforward. Recurring analytics work often depends on someone remembering to review activation, retention, conversion or errors. Scheduling an agent could turn those reviews into an ongoing process, reducing the gap between collecting data and acting on it.

The update also builds on PostHog’s broader move toward agent-driven work. The company previously opened access to PostHog Desktop with multiplayer spaces and autonomous workflows. Scouts appear to extend that direction by adding persistence and scheduled execution.

The more unusual component is a global scratchpad shared by scouts. According to D’Ambra, agents use it to learn over time rather than treating every scheduled run as an isolated task. The scouts reportedly refer to one another as “sibling scout” when using this shared context.

A common scratchpad could help agents preserve findings, conventions and previous decisions. For example, one scout’s discovery about how a team defines an active customer might be available to another scout conducting a related analysis later. That could reduce duplicated investigation and make recurring outputs more consistent.

This approach is particularly relevant alongside PostHog’s work on a semantic layer that gives analytics queries a shared business vocabulary. A semantic layer standardizes what metrics mean; a shared agent scratchpad can retain operational context about how those metrics have been interpreted or used.

It also introduces governance questions. Teams will need clarity about what scouts can write to the shared memory, how incorrect conclusions are corrected, which users can inspect or edit it, and whether sensitive information can move between tasks. PostHog has not yet published documentation addressing those controls.

For now, scouts should be treated as an early product development rather than a fully specified automation system. The announcement confirms three capabilities:

• Agents can run on a schedule.
• They can take actions on a user’s behalf.
• Multiple scouts share a global scratchpad and learn over time.

What remains unknown is equally important. PostHog has not provided setup instructions, supported actions, access requirements, usage limits or a rollout schedule. Those details will determine whether scouts become a practical tool for routine product operations or remain primarily an experimental agent interface.

What is PostHog scouts?

PostHog scouts is a feature comprising automated agents that can run on a schedule and act on a user’s behalf. The agents share a global scratchpad, which allows them to retain and reuse knowledge over time.

What can PostHog scouts do?

PostHog has confirmed that scouts can perform scheduled actions, but it has not published a complete list of supported tasks. Specific examples, integrations and configuration options were not included in the initial disclosure.

When will PostHog scouts be available, and what will they cost?

PostHog has not announced general availability, pricing or plan requirements for scouts. Users will need to wait for formal documentation or a changelog entry before determining access and usage limits.