# What is PostHog? (Official Demo)

## Метаданные

- **Канал:** PostHog
- **YouTube:** https://www.youtube.com/watch?v=1FZji2L-LmM
- **Дата:** 23.04.2026
- **Длительность:** 8:50
- **Просмотры:** 2,230
- **Источник:** https://ekstraktznaniy.ru/video/47553

## Описание

In this PostHog demo, we explain what PostHog does and how to get started. Got a question? Leave a comment. 

🪄 Get started with npx @posthog/wizard
🌐 Our website: https://posthog.com/
❓ Demo playlist: https://go.posthog.com/demos
🆕 Changelog: https://posthog.com/changelog

Chapters:
0:00 – Posthog explained in 90 seconds
1:21 – MCP analysis with PostHog AI
2:12 – Investigating website conversion
3:52 – Summarizing user feedback
4:36 – PostHog AI + SQL Editor = Power
6:46 – Even more demos
7:30 – Getting started
7:58 – The second nostril isn't ceremonial

## Транскрипт

### Posthog explained in 90 seconds []

We built Posthog to help developers and product teams answer the only question that matters. What should I build next? Answering this question used to be hard. You had to integrate half a dozen different tools, do a bunch of data engineering. It was slow, expensive, and error-prone. We make it easy by putting all the tools and data you need in one place and giving you a powerful AI agent to help you make sense of it all. Here's how it works. You send us events from your product using our front and back end SDKs. Which events? — Honestly, whatever matters to you. We have tools for every type of product signal. Product analytics to analyze conversion and retention, error tracking to capture code exceptions and resolve issues, LLM analytics for monitoring and debugging AI products. Then you can build powerful feedback loops, test and measure new features with experiments and feature flags, watch how users respond with session replay, and ask them for direct feedback in surveys. Behind the scenes, we handle identifying your users and build a profile about who they are, how they found you, and how they use your product. We also join that production data of any third-party sources like Stripe and HubSpot inside our fully managed data warehouse. This gives you a single source of truth for your products and users that you can query using our pre-baked insights or our custom SQL editor. And this is the beauty of Posthog. It's the only platform with all that context in one place. So, figuring out who your users are and what they need is easy. And if you need help, you can just ask Posthog AI. An AI assistant so good, our users literally won't shut up about it. Let me show you why. I'm going to demo

### MCP analysis with PostHog AI [1:21]

three ways you can use Posthog AI every day. The first is a classic chat with your data scenario. I keep hearing how great the Posthog MCP is. So, I'm going to ask Posthog AI to show me the trend for Posthog MCP usage. It searches our event taxonomy and correctly identifies the events it needs. It creates an insight for me and produces a short summary noting that MCP usage has increased by six times in just three months. It also suggests completing a deeper breakdown, and I ask it to create a report that I can share. It creates a five-point plan to check the event properties for breakdown options, create some insights, and assemble a report as a notebook. The final result is a Posthog MCP usage report for Q1 2026 showing the growth in MCP usage and some deeper analysis that explains how people are using it. You

### Investigating website conversion [2:12]

can also use Posthog AI to analyze and edit existing insights you've created. Here, I'm looking at a funnel insight that shows the historical conversion rate of users who visit our website and go on to sign up to Posthog. James Hawkins, Posthog co-founder and pineapple on pizza denier, has asked me to report on the impact of our website redesign from September. — Hello, Andy. It is me, James Hawkins, actual co-CEO of Posthog. Hey, listen, Andy. Uh our super important investors want to know the impact of our new website. Uh so, can can you help me with that? You are the most handsome and coolest person I know. Only you can do Who wrote this? — What is this? spend hours doing this, but I'm going to ask Posthog AI instead and take the credit. I open the Posthog AI tab and ask it to analyze the conversion trend, giving it the context about the relaunch in the prompt. I also add this insight to Posthog AI's context window, and you can add multiple insights, events, dashboards, and so much more if you need to. Posthog AI then extracts the conversion rate for the days before and after the relaunch and produces a summary showing how conversion has increased from 1. 75% before the launch to 3. 6% after the launch. This summary is just what I needed. I sent it to James, and he was delighted. Wow. Wow, Andy. That was just what I needed. Thanks. You're super intelligent and your rugged good looks are astounding. No. I'm done. This is stupid. No one's going to think I'm actually James. What do we do?

### Summarizing user feedback [3:52]

Next, let's look at how Posthog AI can help you figure out what users want. We run net promoter score surveys for all our apps where we ask, "How likely are you to recommend it to someone and how can we improve? " I'm looking at the NPS survey results for the surveys app, and I see that we have an NPS score of 55, which is great, but I'm interested in the feedback users are giving us. I could read it all, but I know Posthog AI will do it better and faster. So, I click the analyze responses button. This opens the Posthog AI tab with a predefined prompt to analyze user responses. It generates a summary noting the top issues, what users want, and organizing those actual insights by priority. This is very quick and easy, and I'm going to send this report to the team so they can consider what to do next.

### PostHog AI + SQL Editor = Power [4:36]

For my last trick, Illusion, Michael. I'm going to show you how Posthog AI can query data in Posthog's data warehouse using our SQL editor. You can import and join data in our data warehouse from dozens of sources like Postgres, Snowflake, Google Ads, and Stripe and query it alongside production data you collect using Posthog. I'm going to use our SQL editor to query the performance of our Google Ad campaigns, but I have a confession to make. I have no idea how our Google Ads data is structured. I don't even know how to write SQL, but that's fine because I have Posthog AI. I open a new SQL query and ask Posthog AI to write a query breaking down the Google Ads spend by campaign. It reads the data warehouse schema to figure out which tables have the info it needs, creates a SQL insight for me, and produces a short summary of what it's found. It's pulled together data for total spend, clicks, impressions, conversions, and cost per click, which is great, but there's one more cool thing I can do here. I can ask Posthog AI to estimate the revenue impact of these campaigns. With one simple prompt, Posthog AI fetches invoice and customer data from Stripe, joins that with referral data from Posthog, and then matches that with the campaign data from Google Ads. It updates the query, and in just a few minutes, me, who is in fact two kids in a trench coat who don't know how to write SQL, has completed an in-depth analysis of our ad spend. If I can do this, imagine what an actually competent person could achieve using Posthog AI. I like business. Transactions. That's just a small taste of how you can use Posthog AI, but there's so much more I could show you. To learn more, you can visit our changelog at posthog. com/changelog to see what we've shipped recently. Of course, none of this is possible without the built-in apps that make Posthog such a powerful tool for developers and product teams. Whether you're a developer trying to debug problems and ship new features, a product manager profiling user behavior and feedback, or a data analyst investigating business outcomes, Posthog has everything you need to build a successful product. It would take too long to cover every use case for Posthog in a single video, so we built a

### Even more demos [6:46]

playlist of demos to show you how we at Posthog use it every day. We have Michael from the Posthog AI team explaining how they use Posthog's LLM analytics to analyze usage, compare costs by model, monitor and respond to latency and errors, and experiment with prompts using our prompt playground. We have Ross from the batch exports team showing how they use error tracking to identify spikes in exception events, investigate root causes using stack traces and session replay, and set up alerts for critical errors. And there's Abigail from the support team who explains how they use the Posthog data warehouse and our SQL editor to analyze and report on support trends using data imported from Zendesk. See the link in the description for more.

### Getting started [7:30]

It's incredibly easy to get started with Posthog. One command sets up a fully customized Posthog project. Our AI wizard configures the SDKs, defines custom events with properties, and builds a dashboard for your product. It even installs the Posthog MCP. Just run it in your terminal. And remember that Posthog is free to get started. You don't need a card, and our free tiers are famously generous. Thanks for watching. —

### The second nostril isn't ceremonial [7:58]

— By the beard of the king, Steve the horse, have you ever seen anything more surprising in all your many horse years than this nostril-tooting recorders? Well, yes, that one log was pretty surprising, but use your horse ears, Steve the horse, for a truly majestic sound. I had long believed the second nostril merely ceremonial, but here now she proves me wrong. I dare say the only thing more surprising than this bard's nasally melody is how incredibly easy it is to install Posthog using their very own wizard. A real-life wizard, Steve the horse. Can you believe it? And how this installation tool is no conjurer of cheap tricks. Yeah.