PostHog Launches Self-Driving Mode With AI Scouts That Draft Pull Requests.

PostHog announced on LinkedIn that its open-source analytics platform can now operate in self-driving mode, with AI agents that automatically detect product issues and draft code fixes. The feature entered open beta today and can be installed via a single CLI command.

The system relies on background agents PostHog calls "scouts" that continuously analyze product data from connected tools and PostHog apps, including error tracking capabilities the company expanded earlier this year. These scouts surface problems as reports in a centralized PostHog Inbox, where issues are clustered and ranked by priority.

What sets the feature apart is the final step: each report includes a draft pull request generated in a sandbox environment. Developers can review the proposed fix and merge it if approved, or reject it outright. PostHog positions this as a safe middle ground between fully autonomous code changes and purely diagnostic tools that stop at identifying problems.

Developers can set up self-driving mode by running npx @posthog/wizard@latest self-driving in their terminal. The feature integrates with existing PostHog installations and works alongside the platform's analytics, session replay, and experimentation tools.

PostHog offers three ways to interact with self-driving mode. The primary interface lives in the web app, where teams can triage issues and review PRs. The system also integrates with Slack, allowing teams to receive alerts and approve fixes without leaving their chat workspace.

The third option uses the Model Context Protocol (MCP), an emerging standard for connecting AI systems to data sources. This approach lets developers build custom workflows around the scout system or integrate it with other AI tools.

The self-driving feature follows PostHog's broader investment in AI-assisted workflows. The company built a setup wizard that uses an AI agent to automate onboarding, handling tasks like installing SDKs and configuring tracking without manual intervention.

By releasing self-driving mode in open beta, PostHog can gather feedback on how teams actually use autonomous issue detection and whether developers trust AI-generated fixes enough to merge them. The sandbox approach addresses the obvious risk: no one wants an agent pushing untested code directly to production.

The feature assumes teams already use PostHog for analytics and have connected the relevant data sources. For organizations running PostHog primarily as a passive monitoring tool, self-driving mode represents a shift toward active, automated remediation.

PostHog has not disclosed pricing for self-driving capabilities or whether they will remain available across all tiers once the beta period ends. The company's open-source model typically means core features stay accessible to self-hosted users, though some advanced functionality moves to paid plans.

Developers interested in testing the feature can install it now via the CLI wizard. PostHog's official site provides additional documentation on configuring scouts and managing the inbox workflow.