Why the Future of Software is AI and Human Collaboration | Steven Fabre, Co-founder and CEO, Liveblocks

Thanks for tuning in to The Peel; I’m Turner Novak from Banana Capital. Today I’m joined by Steven Fabre of Liveblocks to talk about what people get wrong with AI-native software, treating AI as more than a sidecar, how enterprises are actually buying, his shift from designer to CEO, and rebuilding after a cofounder left, plus a quick plug to browse the archive and come back next week for Jason Calacanis.

Liveblocks is collaboration infrastructure for product teams, with drop-in components for comments, presence, notifications, and real-time editing so you can build Figma or Google Docs–style experiences fast. Our big push is AI as a true collaborator you can mention anywhere in the workflow so it can act on context and do work like a teammate.

How does that integration actually work, especially across different models and agents?

Developers use our React components and APIs to embed AI in comments, chats, notifications, or even document editing, then point it at whatever model or agent stack they prefer while we handle orchestration. Think of it like Stripe for collaboration UX: we provide the interface that sits above your chosen models and agents.

Beyond a bot in the corner, how are you seeing AI show up in products in a useful way?

Bolting on a worse version of ChatGPT rarely lands; the better pattern is letting people summon AI right where work happens so it behaves like a colleague. For example, @mention an AI engineer in Slack to file an issue, make a fix, and open a PR, similar to what tools like Linear, Vercel, and Cursor are enabling inside their own flows.

So the UI may not flip to all-chat; it meets users where they already work?

I expect familiar collaboration patterns to endure, with chat and even voice helping in places without replacing everything. We’ll likely see more discovery and even buying inside chats, which turns into an AI-era SEO problem more than a wholesale UX reset.

There’s a wave of GEO startups chasing visibility in LLM answers, but distribution power still sits with the platforms.

ChatGPT is already a top traffic source for us and a strong onboarding funnel when it surfaces plug-and-play steps inside the chat. Given how much training data comes from Reddit, brands are seeding authentic discussions there because those threads influence model responses.

You’ve crushed launches before; what’s your playbook for getting attention without the fluff?

We overprepare with tight visuals and a great demo, kickstart momentum across time zones right after midnight Pacific, and do a lot of direct outreach to champions. Beyond leads, these launches create internal urgency, clear deadlines, and real moments to celebrate.

How did the product evolve as you moved upmarket, and what did you learn about enterprise sales?

I learned by doing with an advisor on my calls, which sharpened pricing, positioning, and the roadmap through real objections. Moving upmarket lifted ACVs but made us feel less approachable to early-stage teams, so we’re rebalancing toward stronger self-serve while keeping enterprise strong.

Big contracts are moving fast due to clear ROI and “innovation” budgets, but it has to tie to real outcomes.

Exactly, we want durable value, not science projects, and we’re aiming for efficiency so growth isn’t dependent on a huge sales army. The classic triple–double cadence is harder to justify if you can drive more with product-led motion.

As a designer turned CEO, why is design central here, and how did Liveblocks originate?

Design is structured problem solving, and running a company feels like designing a system that produces outcomes. At InVision we struggled to build real-time, multiplayer creative tools, which led my cofounder Guillaume and me to focus on the underlying collaboration infrastructure instead.

What’s the backstory with Guillaume, and how did that shift you to infra?

We bonded as two folks from the same tiny region in France and paired closely on a browser-based presentation tool until we realized the real pain was the real-time stack. That’s when we decided to build the building blocks everyone else needed too.

When he stepped away, how did you handle it and keep momentum?

It was the hardest chapter, but we kept communication open, aligned stakeholders on terms, stayed friends, and channeled it into focus so the team stepped up and revenue grew several times. Hiring the right head of engineering took nearly a year, and we finally landed someone with deep AI and real-time experience.

You also built a remote-first culture early; what actually makes it work?

We insist on daily time-zone overlap, meet in person at least once a year with optional project meetups, document decisions well, and hire experienced people who thrive remotely. Not everyone is a fit for remote, and that hiring bar matters more than any tool.

Quick hoops detour: underrated skills or players, and your own path into tech?

Versatile bigs and relentless defenders are priceless, and a knee injury in Sydney pushed me from restaurant work into freelancing websites, which kickstarted my product career. That pressure taught me to build, sell, and ship fast.

This was a blast; catch last week’s chat with Zayd Ramzan of Flex on building an AI-native private bank for the middle market, and come back next week for Jason Calacanis from All-In. If you enjoyed this, subscribe, leave a comment, maybe name your next AI feature after me, and grab my newsletter The Split for episodes, transcripts, and drafts in your inbox.