Skip to content
YC Root AccessYC Root Access

PostHog: Pivots Were The Real Lesson In Building A Startup

In this fireside from Startup School Paris, PostHog CEO James Hawkins talks about surviving five failed ideas, why going open-source nearly trapped the company in enterprise on-prem debugging hell, and how a hackathon side project by one engineer turned into a multi-product strategy neither founder had planned. Apply to Y Combinator: https://www.ycombinator.com/apply Work at a startup: https://www.ycombinator.com/jobs Chapters: 00:00 — Intro 00:46 — The Rule: Sleep on It Once 02:13 — Why Pivots Were the Real Education 02:31 — Birth of Open Source Analytics 05:14 — How They Knew This Idea Was Different 05:49 — Competing Against Amplitude & Heap 07:17 — The On-Prem Nightmare 08:57 — Going Cloud & Killing the Friction 10:33 — The Engineer Who Forced Multi-Product 13:06 — When Ambition Exploded 15:45 — The Shift to Self-Driving Software 17:18 — The Agent That Beat All Other UX 19:29 — Fixing Bugs While You Sleep 21:15 — Half of PRs Are Getting Merged 23:02 — Betting on Where the Puck Is Going 24:27 — Elephant Carpaccio 29:06 — The Founder Who Became a PM 31:49 — Why Going Wild Gets Easier 34:41 — Free Up Everything That Isn't This

James Hawkinsguest
Aug 19, 202638mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

PostHog’s pivots led from open-source analytics to self-driving software

  1. PostHog’s most important early lesson wasn’t a single idea but learning to pivot fast through repeated build-and-test cycles until something felt both fun and commercially viable.
  2. The company’s initial open-source, self-hosted analytics thesis generated demand but created an on-prem support and enterprise-sales gravity that pushed them toward a cloud-first, self-serve, low-friction model.
  3. PostHog evolved into a multi-product suite largely through shipping momentum and competitive pressure, leveraging shared data context and open-source components to outpace point-solution rivals.
  4. In 2025, PostHog reframed its platform around an AI agent that became the primary user experience, enabling “self-driving” workflows that prioritize and even fix issues via automated pull requests.
  5. Hawkins describes AI transformation as founder-led and operationally extreme: freeing time, restructuring responsibilities, and managing customer and internal change while gradually migrating from traditional UI to agent-first experiences.

IDEAS WORTH REMEMBERING

5 ideas

Fast pivots can be a deliberate learning machine, not a failure mode.

PostHog cycled through six startup ideas in ~7–8 months, using a strict rule: if it still felt wrong after sleeping on it once, they restarted immediately. The goal wasn’t to perfect early strategy but to learn quickly by shipping and watching real user behavior.

Self-hosted “enterprise-friendly” positioning can quietly force you into enterprise sales and support.

They differentiated initially via open source + self-hosting (a GitLab-vs-GitHub framing), but discovered the operational reality: supporting on-prem meant effectively debugging customers’ infrastructure (often Kubernetes) and pulling them into slow, conservative enterprise motion.

Owning the product experience (cloud) can be more valuable than preserving the original open-source go-to-market thesis.

They moved to a self-serve cloud product with a big free tier and low friction (pricing transparency, easy onboarding, usage-based thinking). This sacrificed near-term value capture but expanded adoption and let PostHog control reliability and perceptions of “bugginess.”

Multi-product can emerge organically—and become a competitive moat—when shipping velocity and shared data/infrastructure compound.

Multi-product expansion wasn’t a grand master plan; it began when an engineer shipped Session Replay despite founders thinking it was a bad idea. PostHog then leaned into becoming a suite because competitors were point solutions and they could ship fast by leveraging open-source infrastructure like ClickHouse.

Agentic UX can simplify a complex platform by replacing navigation with iterative problem-solving.

As the product surface area grew, customers complained it was overwhelming; AI became a “saving grace” by turning an agent into the main UX. Once the agent could iterate in a query loop (try → evaluate → try again), it moved from mediocre to “better than a human,” driving the majority of positive mentions.

WORDS WORTH SAVING

5 quotes

Our only rule was if we felt it was something wasn't, wasn't working, we would make ourselves sleep on it for one night, and then if we still felt like that the next day, we would just start afresh.

James Hawkins

The reality of running this business was, like it's growing, growing pretty well, and we're basically debugging other people's Kubernetes over Slack.

James Hawkins

This is kinda my grand theory is like, well, it's better to solve the problem than to surface the problem.

James Hawkins

I would say competitively as well, like I think, um, we're like, "Oh yeah, why not do the same Claude or something?" We're like, well, if we or we know before a human does there's a problem, so we can ship the fix, um, before Claude can get there.

James Hawkins

62%, I think, of pull requests that come through Slack are not from software engineers at PostHog anymore. It's from, like, the marketing team, for example-

James Hawkins

Rapid pivoting discipline (“sleep on it once”)Open source vs cloud control of UXOn-prem/self-hosted operational burdenSelf-serve pricing, free tier, friction reductionCompeting with Amplitude/Heap/Mixpanel via suite strategySession Replay and emergent multi-product directionAgentic UX, iterative query loops, self-driving mode

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.