PostHog AI example prompts
Contents
PostHog AI can be used in many ways. Because it works directly with your project's data, its responses and actions are unique to your setup. There is no single "correct" way of using PostHog AI.
As a starting point, here are some example chat prompts you can use to explore what's possible with PostHog AI.
LLMs
PostHog AI's responses and findings highly depend on the product data, context, and prompts you provide it. As with all LLM-based systems, its non-deterministic nature means the same prompt can potentially produce different outputs.
Product analytics
Create an insight showing daily active users for the past weekBuild a dashboard for onboarding retentionWhat changed across my product this week?
Web analytics
Create a dashboard showing web traffic by source and campaignWhich landing pages have the highest bounce rate this week?Find the most common paths users take after visiting the blog
Session replay
Find sessions where users dropped off during checkoutSummarize session replays where users got stuckShow recordings for users who triggered the rage click event
Data warehouse
List all available tables and their columnsWrite a query to join event data with user propertiesFix the syntax for the query I have open in the SQL editor
Feature flags
Create a new multivariate feature flag for dark modeShow me which flags are currently active in production
Experiments
Set up an A/B test with a 70/30 split for a new red button on the homepageSummarize experiment results for my latest feature rollout
Error tracking
Find the most common frontend errors this weekWhich errors impact user sign-ups the most?Summarize new issues introduced after the latest release
Data pipelines
Create a Hog function that filters out internal trafficParse event timestamps into day, week, and month propertiesDrop events where test_mode = true or the environment is staging