The AI platform for engineers
Debug products. Ship features faster.
With all user and product data in one stack.
Overview
PostHog AI is deeply connected with your data
Ask it to build insights, write PostHog SQL queries, summarize session recordings, create surveys, set up feature flags – basically handle the grunt work that would normally take 20 minutes of clicking around.
Unlike other agents, PostHog AI can:
- Understand and explore your product data in depth
- Synthesize context across multiple sources
- Create dashboards and data visualizations in PostHog
- Interact natively with the PostHog UI, for example by editing filters
- Generate and edit complex SQL queries
- Answer questions from PostHog's product documentation
Features
Analytics
Product engineering
Communication
PostHog data stack
PostHog AI knows...
Product Analytics
Less digging - more dialogue. PostHog AI lives in your product data. Create insights and dashboards, and generate complex HogQL queries with natural language.
Skills
The 
Build insights from plain English
Describe the data you want to see visualized as an insight or dashboard
Write and explain HogQL
Create complex SQL queries and get an explanation of the logic behind them
Deep dive into your product data
Try research mode for deeper analysis across multiple context sources
Navigate the UI
Edit filters, set breakdowns, build retention curves (and more) with simple prompts
PostHog AI everywhere
PostHog AI, wherever you work
Bring your product data into the tools you already use.
MCP docs
In your coding agent
Plug PostHog into Claude, Cursor, and anything else that speaks MCP.
- Check before you change – "is anyone still on this flag?" before you delete it; "how often is this endpoint hit?" before you rewrite it.
- Size the impact – check who's on a feature flag or how an experiment is trending before you decide to ramp, hold, or roll back.
- Verify the deploy – see whether latency, errors, or conversion actually moved before you ship the next thing.
Slack app docs
In Slack
Mention @PostHog with a data question, quick fix, or chore.
- Pull product data into a thread – conduct research and query data to answer product and customer questions.
- Subscribe to product insights and dashboards – receive a regular digest in your chosen Slack channels.
- Set up Slack notifications via PostHog Data Pipelines – get timely notifications of any event or action in PostHog.
Demos
Try it
Chat with your data
Still building insights manually? Ew. Let PostHog AI help.
"What changed across my product this week?"
PostHog AI connects every data point to see the full picture.
"Find sessions where users got stuck"
PostHog AI can summarize recordings and suggest next steps.
"Build a dashboard for onboarding retention"
PostHog AI builds dashboards and insights from plain English.
"Analyze LLM token usage over the past 7 days"
PostHog AI helps you build better AI products. (How meta.)
"Find the most common user path from the blog"
PostHog AI analyzes web traffic faster than you can say "I miss GA3."
"What's the right SQL syntax for this query?"
PostHog AI writes and debugs SQL queries (so you don’t have to).
"Find the top 10 orgs by number of active users"
PostHog AI joins with external data sources for even more context.
"How do I set up a feature flag?"
PostHog AI is an expert on using PostHog products. (Duh!)
Signed into PostHog?
Click any question to get PostHog AI started.
Advanced modes
Advanced modes
PostHog AI auto-switches basic modes based on your query. Activate advanced modes in the chat for more powerful analysis.
Plan mode
Think twice, ship once
For complex queries without clear solutions, ask PostHog AI to map out the approach. Plot the next five moves before diving in.
1
Clarify the mission
Time period? Segments? Success criteria? The agent asks upfront (it's like a standup, but useful).
2
Approve the battle plan
No need to burn tokens on the wrong approach. See exactly what the agent wants to do. Tweak it. Reject it. Or let it rip.
3
Watch it execute
Once a plan is approved, PostHog AI builds to spec. Dashboards, SQL queries, and insights – delivered as specified (no scope creep).
Videos
How PostHog uses PostHog AI
PostHog AI for you
What PostHog AI can do for [you]
Move fast and break (fewer) things
PostHog AI surfaces issues hiding in session recordings and error logs so you know about issues without users having to report them.
Get a technical co-founder that never sleeps
Query your data in natural language and get insights instantaneously.
See the patterns hiding in product data
PostHog AI analyzes user behavior, surfaces the ‘aha,’ and helps write the playbook for what to build next.
Roadmap
Roadmap
Concept
Ideas we have committed to exploring.
AI API access
A lot of people want to build on our APIs. We'll soon offer a way to let people access our AI via API so they can offer in-built AI capabilities to their in-app analytics.
ReviewHog (PostHog code review)
An AI code reviewer for your GitHub pull requests. Every change is reviewed from multiple perspectives and every finding is validated against your codebase before it reaches the PR. The perspectives and the validator are customizable skills: tune the built-in ones or add your own.
Alpha
In testing with a small group of users.
Nothing in this stage right now.
Beta
Ready to enable and try in PostHog.
Nothing in this stage right now.
Have opinions about what we should build next? Vote on our roadmap.
Pricing
Pricing
Monthly free tier
500 credits/mo
Then starts at
$0.01/credit
Volume discounts
First 500 credits/moFree
500+$0.01/credit
Plans
Free
No credit card required
Paid
Starts at $0/mo
Credits
500/mo
Unlimited
Features
available on both plans except listed below
Slack agent
✗
✓
How credits work
AI credits are based on the underlying token costs, which reflect the effort required to complete your request.
- Simple queries like "What were my daily active users in October?" use very few tokens, and therefore very few credits.
- More complex tasks like analyzing hundreds of session recordings or rewriting a SQL query multiple times use more tokens and consume more credits. While exact usage varies, credit consumption usually scales with complexity – more advanced tasks cost more but can deliver deeper insights and time savings. You’ll always see real-time cost information while using AI features.
PostHog automatically selects the most efficient model for each AI feature. We apply a simple, consistent 20% markup over the underlying LLM provider’s cost: So 1 PostHog AI credit equals $0.008333 of raw inference, and 100 credits cost $1.
Get started
Colors
Foreground
Background
Canvas: 640 × 480
Zoom: 100%
Tool:
Help? Sorry, I can't help with taste! 🎨
Get started today!!
If you we're looking for a sign...(yes, the typo was intentional)
Follow your dreams
Install with AI
in less than 90 seconds
Works in your terminal or with tools like Cursor, Replit, and Bolt.
npx -y @posthog/wizard@latestSupports Next.js, React, React Native Svelte, and Astro
(Probably best to do this from a computer though)
PostHog AI is my passion
New year, new you!
Get started with a new better life
No presenter notes for this slide.