Pawel Cebula

eng @posthog 🦔 • prev. eng @joinodf, founder @MEDIGOcom • powered by 🍜🌮🍗🍺☕+☀️🌊🏄‍♂️🏔️🏂

[PostHog @posthog]

He's been watching your session the whole time.

[PostHog @posthog]

Introducing PostHog Code, the product editor that:

- Understands your product
- Identifies usage patterns
- Triages bugs and errors for you
- Creates PRs to fix them
- Continuously monitors and improves your product

Join the waitlist: https://t.co/pTuwqCzo1I

[PostHog @posthog]

Today we announce an exciting new chapter in PostHog’s journey, our new range of PostHog Heroes action figures.

With PostHog Heroes, our precious youth can finally realize their dreams of one day founding a B2B SaaS company and delivering shareholder value.

We’ve already sold the movie rights to A24 with Timothée Chalamet attached as the lead. We’re hopeful this is the movie that will win him the Oscar he craves.

[james hawkins @james406]

Hello everyone!

We’ve got some exciting news to share. We raised a Series E – $75M at $1.4Bn, led by @peakxvpartners with significant participation from existing investors. We can now proudly upgrade our status from unicorn adjacent to unicorn.

Most will talk about how they feel humbled at this stage and, let’s face it, even the biggest fans of LinkedIn will glaze over. So, I’m going to tell you a story along these lines instead.

I was in San Francisco last week. One evening I’d been at a dinner with a group of other startup founders. Afterwards, I was walking back to my hotel when suddenly, out of the gloomy darkness, probably 100 yards behind me, I heard “JAAAAMES”.

If you’ve ever walked along Mission late at night, this meant I was either about to get assaulted, or a high five. I turned around and saw a guy leaning out of his car window. “OH MY GOD NO WAY! IT’S THE POSTHOG GUY. I LOVE POSTHOG!!”.

I probably wouldn’t have recognized my mom in those conditions. It really struck me that we’ve always felt supported by our users and fans – we’ve been able to do so much of our work with a smile on our faces. We are very lucky. Either that or our billboards with my face on worked.

So, why’d we raise?

We are starting Act 2 of PostHog. We are going deeper into being a devtool, not "just" analytics. We will keep improving our existing tools, and adding more of them to generate better insights into how your products are performing, but we're also going to start working on generating pull requests for you.

Today, we know that engineers like you start work, and immediately wade through escalated support tickets, Slack, email, whatever customers said in the latest call – often only by the afternoon to start working on new features.

We want you to be able to start work, and then immediately see a list of PRs, already made for you automatically based on all the things that popped up since you last logged in.

This means for smaller issues, your workflow gets you straight into merging, closing, unblocking, or editing. It doesn’t stop there, though. Once a feature is released, we add things like event tracking, feature flags, experiments and even messages to be sent for feedback, automatically.

This is likely to overload my inbox, but if you’re interested in hearing more then email me back. We’ll gradually be releasing the very earliest versions of this to keen early adopters for feedback.

On Peak XV

Tim and I first came across Peak XV when we saw our friends at Supabase raise from them, then Arnav (someone who has been working with us closely for years) introduced us to Shailendra. We just clicked.

Choosing Peak XV was about the people. In every meeting with Shailendra, he has come across as genuine, excitable, and hard working. Peak quickly proved themselves a delight to work with, and when we received an offer from them, we felt they were the best way for us to keep building. And that’s all that really matters to our users.

So, on that note, please go back to building your companies!

- James, Tim and our spiky little team at PostHog

[james hawkins @james406]

warning: actually non-meme content...

@timgl and I jumped on some quick calls and ended up raising a $70 million series D led by Stripe.

and of course we decided to announce this with the help of our newly created irl hedgehog Max.

see reply for a link to a post on how we’re going to use the $$$, but here’s the tl;dr version:

- we'll build even more products, like messaging, support, CRM, a lot more
- all the tools + your data + Max AI = shareholder value

[Pawel Cebula @pawelceb]

@michalmalewicz @thekitze +1, not sure what styles you're after but here are some of my favorites, you can order from here: https://t.co/Znqa2xpp63

[PostHog @posthog]

Our error tracking beta is now... no longer a beta!

Like all our products, there's a generous free tier – 100k exceptions every month. After that, pricing is usage-based and very affordable.

It does all the things you'd expect an error tracking tool to do, like:

- Notifications of new issues by email, Slack, or webhook
- Triage issues by assigning to individuals or groups
- Stack traces, so you can see the code context associated with an exception.

But all these things are made 10x better because it integrates with all our other products, such as:

- Session replays, so you see exactly how and when things break.
- Feature flags, so you can test fixes by rolling out code changes only to affected users.
- Product analytics, so you can graph `$exception` events and analyze their impact.

Head to the docs to learn more: https://t.co/nvifCDc5Qe

[Pawel Cebula @pawelceb]

If you're considering moving to the US, check out this collection of breakout startups sponsoring visas!

[weisser @julianweisser]

New experiment for ODF23: pay-what-you-want

ODF22 was our best cohort (NPS/economics) but we wanted to try something new.

Not changing: no equity to join, no right to invest. We're not a VC's dealflow hack.

We help more people start the best companies together. Apply!

[Pawel Cebula @pawelceb]

@richardpoelderl Haven't tried Supermaven but have been using Cursor for the past month and it feels miles ahead. Takes literally 10 seconds to switch from VSCode.

[ODF @joinodf]

Announcing: ODF’s Top Companies of 2024

In 5 years, ODF founders have started 1k+ startups worth billions.

Here’s our list of companies that made major progress in the last year, ranging from pre-seed to series D.

https://t.co/ZIngNV68S0

[Pawel Cebula @pawelceb]

Lorenzo is a master at wrangling complexity and making things simple for your team and users, and he really just cares - had a chance to see his amazing level of dedication in action every single day over the past couple of years.

[Lorenzo Castro @Lorenzo__CB]

Today @vladoustinov and I are launching Railblocks 🚀

We partner with ambitious founders building ops-heavy business to help them design, build and grow their own internal Operating System.

Quick 🧵

[Pawel Cebula @pawelceb]

@andrew__rea @rhobusiness @justinwolz @TheNFTCPA @skbjml Which chicken parts were highest/lowest in demand?

[Pawel Cebula @pawelceb]

@Nils_Reimers @cohere Will definitely try max scaling though!

[Pawel Cebula @pawelceb]

We've been happy users of @cohere rerank EN v2 for keyword and vector search results. Started testing v3 recently - it does better in bringing most relevant results to the top but the distribution of relevance scores is wider (hard to define cutoff). Anyone else noticed this?

[Pawel Cebula @pawelceb]

@Nils_Reimers @cohere Agreed cutoffs are always tricky. v2 seemed a bit more consistent in terms of ranges - average set based on a sample of "just about irrelevant" was somewhat meaningful. With v3 the ranges seem a lot more all over the place. Relative rankings are definitely better than v2 though.

[Pawel Cebula @pawelceb]

@CharlesLiu9 @zlwaterfield +1 on the book! I've read first two books so far. I've also watched the first episode of the Chinese produced series - so far it feels like a much better/closer depiction of the book. It's 30 episodes long though (covering first book, like the Netflix series)

[Pawel Cebula @pawelceb]

@pitdesi @alexkehr I'd argue that Foursquare pulled it off at some point in major geos. By far most reliable ratings in this space. AFAIK big factor in the score was frequency of returning customers relative to other restaurants in the area etc. Such a shame that they didn't last as a consumer app

[Pawel Cebula @pawelceb]

@nonmayorpete Had this issue too and uploading new profile photo for the GPT helped, I've not touched it before

[Pawel Cebula @pawelceb]

@labenz It could be some issue on my part with training data format in a chat completion API, though I've followed the docs and tested a couple of different formats and this was the best I could get. It might also be that somehow the fine tuned GPT3.5 doesn't handle classification well