Sunil is the founder of Partykit - a platform for building real -time collaborative applications like a Google Docs or Figma, or even just a multiplayer game.
Partykit was backed by Sequoia and a bunch of awesome angels (incl. me) and was recently acquired by Cloudflare. Sunil has spent over a decade as an engineer, community organizer and worked on popular OSS projects like React. I discussed a bunch of topics with him from the gritty details of his acquisition to the grim developer tools funding landscape to his issues with developer relations and views on open source.
Below are the highlights from our conversation:
Why did he sell his company to Cloudflare - despite having enough cash in bank, Sequoia as his lead and inbound VC interest for the next round?
Sunil says “people's expectations of what application should be has just changed in the last two years. You can call it hype, but the LLM thing is real. Everyone's fundamentally rethinking everything. So in the middle of that, to be building technically a non -AI company and the combination of these five things: open source, B2B, SaaS, Infra VC backed was a very challenging combination”
Pai adds “I thought this is probably the best thing for the product and the platform to survive and grow.” He didn’t shop around because according to him “Partykit’s platform and the roadmap is so very dependent on this tech that no one else has” and Cloudflare gave him the freedom to allow building it out without any serious oversight.
“And I'll be honest, being able to focus on building that without worrying about enterprise sales in 2024 was very, appealing as an option.”
How did Sequoia feel about this?
Pai says “my lead was Sequoia, who I had a great relationship with. Still do. They were like fully supportive of it and they believe in the founder. And if the founder says, hey, like, I don't think this is the right thing to do. They can't really force you to keep going. I was feeling bad that I'm not able to like return all the money in the pot and stuff like. And they were like, dude, like you're like a pre seed investment. This is it's absolutely fine. It's not a material difference for us. And when you do make another startup in the future, please let us know.
Future of real time multiplayer web
Advances in stateful serverless, push driven architectures, tools like Partykit, Cloudflare’s durable objects, local-first, CRDTs and synchronisation across devices will help push the limits of multiplayer web. And multiplayer will not be limited to humans - the rise of AI Agents developing context will lead to a whole class of applications and experiences being developed that were either not possible or just in the realm of billion dollar companies.
Large Company OSS motivations
Large companies who have been successful custodians of OSS have a top-down strategic mandate to do so. For example Meta’s focus on React, Pytorch, LLAMA pays off via incredible talent acquisition (which it knows how to monetize). Apple on the other hand - doesn’t have an OSS mandate and hence developers wouldn’t see it as a great custodian of OSS software.
Startups and Open Source
VC backed startups need to really think hard before deciding to open source their core technology. Don’t just use it for marketing and “getting adoption” - think a few layers deeper. It either has to have strategic or commercial value for you or it has to be almost about commoditizing your compliments. Also - its less about being secretive and more about it taking a lot of effort - often upwards of 1 day a week. And if you do go ahead - think long and hard about about the commercial licensing. Don’t worry about your “open source” cred. In today’s world, people with budgets (including technical folks) don’t care about it.
Why is so hard to build developer tools in 2024?
Devtools as a path to the riches has gotten a lot narrower. Especially if it’s open source because you need 2 hits now - the OSS project and the commercial offering. With IPO, M&A window and awful multiples - and developers still a miserly lot (accelerated by the JavaScript boom) that don’t pay - developer tools aren’t that exciting anymore. In 2024, you need to be have a compelling value proposition that will make companies pay for either compute, storage or inference (the typical revenue models) and pay a number a lot more above what your costs are.
The end of ZIRP DevRel
Very few companies know how to measure devrel’s impact (Vercel & Lee Robinson is really good at this). Either have devrel report to marketing (in which case their metrics should be aligned to pipeline metrics) or if they’re reporting to product - then hold them accountable to product metrics / bringing in new feature requests. Explosion of devrel post 2018 destroyed the conference and meetup scene by focusing more on product pitches rather than talk about taking feedback on ambitious experiments.
Getting love from developers
Best way is to keep showing your work working in real conditions and keep talking to your customers/users. Get your landing page to be very clear on what your product does (and ,ideally, doesn’t do) and the specific problem it can solve for your users. Pai adds perhaps that’s what authenticity is.
Note: If you’re an Indian operator in Europe thinking of starting up or an Indian founder in Europe - I’d love to chat and learn of your journey
Multiplayer web, death of devtools, devrel and OSS startups with Sunil Pai of Partykit / Cloudflare