Use Logs in PostHog Web

The PostHog web app is home base for Logs. It's where you search log records, narrow them down with facets, mine patterns across services, and set up the alerts that tell you when something changes.

Most investigations start the same way: filter to a service and severity, find the log line that matters, then pivot to the session replay, person, or error attached to it.

What you can do here

• Search and filter – narrow logs with the facet rail, the filter bar, or free-text search, and save a filter combination as a view you can switch back to.
• Mine patterns – group similar log lines into templates to spot noisy logging, new error shapes, and changes in your traffic.
• Set up alerts – get notified when the volume of logs matching a filter crosses a threshold, with noise-reduction settings so brief spikes don't wake anyone up.
• Explain a log with PostHog AI – open a log record and have PostHog AI write up what it means and what probably caused it.

Explain a log with PostHog AI

Open any log record and switch to the Explore with AI tab, and PostHog AI explains that line: a severity assessment, ranked probable causes, suggested next steps, and the fields worth your attention. It's handy when the log came from a service you don't own.

This is a feature of the log detail view, not an assistant you can hold a conversation with. It explains the one record you have open – it doesn't search your logs and it can't act on them. Finding the right record is still the job of filters, patterns, and free-text search, or of your coding agent over MCP.

See Explain a log with PostHog AI for what it returns and what it reads.

Configure Logs for your project

A few things are project setup rather than day-to-day investigation, so they live outside the logs viewer.

PII scrubbing is a toggle in Settings → Environment → Logs. It redacts emails, bearer tokens, and secret keys at ingestion, before anything is written to storage.

Sampling and drop rules live under Logs in the app. They decide what share of high-volume, low-value log records you keep, which is the main lever you have on cost. Set these up once when you onboard a noisy service, not while you're debugging.

Related

• Get your first logs flowing in Start here.
• Query logs from your editor over MCP.
• Script alerts and exports with the Logs API.
• Make your logs worth searching with logging best practices.