Recently, we brought together two OpCo portfolio founders for a candid conversation about what it takes to build an AI-native company. They’re at different stages of their company-building journeys -
Kieran Snyder
(Co-Founder, Chief Scientist Emeritus, and former CEO of
Textio
) has been building with AI for nearly a decade, and
Jerry Zhou
(Co-Founder and CEO of
Supio
) is fresh off raising a $25M Series A - so there was a lot to unpack. One thing they have in common? Telling it like it is. Here are their spiciest takes and most valuable insights - hear it for yourself in the video below. Special thanks to our friends at
Bonfire Ventures ??
,
Ascend
, and
Fenwick & West
for partnering with us!
Building Something Different When Everyone Has the Same Tools
The elephant in the room: how do you build something unique when everyone's using the same foundation models? Both founders had strong opinions on this:
- Vertical > Horizontal: Kieran dropped this truth bomb: if you’re using the same foundation models that everyone else is using… which you probably are… you’re going to compete better if you’re able to build super tailored proprietary data sets, and your app will work a lot better for your customers. Go deep into specific industries.
- Data Moats Still Matter: But not in the way you might think. It's not about hoarding massive datasets before you launch. Instead, focus on building your product so it creates valuable data loops from day one. Every user interaction should make your product smarter in ways that matter for your specific use case.
- Customer Problems > AI Solutions: Jerry's take was refreshingly straightforward - stop obsessing about AI and start obsessing about customer problems. He shared how Supio succeeded by deeply understanding law firms' workflows first, then figuring out where AI could make those workflows dramatically better.
The Price Is Right... Or Is It?
Both founders got real about pricing strategies:
- Start Simple: Jerry started with basic per-seat pricing just to prove people would pay anything at all. The key? Just get those first few customers paying something.
- Evolution is Normal: As Supio grew, they completely transformed their pricing model. They switched from flat monthly fees to usage-based pricing because that's how law firms actually think about costs. The result? Their ACV more than doubled.?
- The AI Premium Question: Kieran shared an interesting perspective on whether to charge extra for AI features. While some big companies are bundling AI features "for free" (because they can afford to), smaller companies need to be strategic about compute costs. Textio's solution? Bake AI into everything but be smart about when and how you actually use it behind the scenes.
The Talent Challenge
Everyone's worried about competing with Big Tech for AI talent. Our founders had some real talk on this:
- Don't Play Their Game: Kieran was blunt - if someone's primary motivation is a big paycheck and stability, they probably aren't right for your startup anyway. Look for people excited about having outsized impact and accelerated growth opportunities. “The only way to get great talent is to get those who want the high risk high reward of being at a startup.”
- Engineers > Data Scientists: Jerry's hot take - it’s easier to train engineers to learn data science than it is to train data scientists to learn engineering.
- Technical Credibility Matters: Having technical founders or leaders who can speak the language makes a huge difference in recruiting. Kieran has a PhD in natural language processing so it was easy to tell her story to technologists and get them excited. Engineers want to know they'll be building cool stuff with people who get it.
Surprising Success Strategies
Some unexpected wisdom emerged:
- Turn Weaknesses into Features: Both founders emphasized this approach. When asked how Textio itself makes sure its models are bias-free, Kieran notes that… they don’t! The models are designed to create a biased outcome. And instead of trying to eliminate AI hallucinations entirely at the foundation model level, Supio built features around hallucinations, turning a potential weakness into a selling point.
- Speed > Perfection: In AI companies, you can build impressive demos fast. Use this superpower! But be ready to pivot quickly if something isn't working. As Jerry noted, sometimes someone can solve in 5 minutes what another team spent 6 months struggling with. Kieran said the only thing she’s seen work is to reward speed and hold people accountable to it.?
- Challenge but Love Your Customers: Jerry shared how he used his technical skills to "delight people with code" even when he didn't fully understand their industry yet. The key was bringing his authentic expertise while being eager to learn their world.
The Bottom Line
Building an AI company in 2024 isn't actually about AI - it's about solving real problems in new ways. The founders who succeed aren't the ones obsessing over having the best models or the biggest datasets. They're the ones who deeply understand their customers' problems and use AI as just one tool in their arsenal to solve those problems in delightful ways.
Whether you're just starting out or scaling up, focus on bringing authentic value to a specific vertical, price in ways that align with how your customers already think about value, and build teams that are excited about the actual technology and the startup journey. The rest will follow.
