AI startups face a different set of challenges than typical SaaS companies. That was the message from Rudina Seseri, founder and managing partner of Glasswing Ventures, at the TechCrunch Early Stage event in Boston last week.
Seseri made it clear that just because you connect to some AI API doesn't make you an AI company. “AI-native doesn't mean his AI company slaps on a glossy wrapper that calls OpenAI or Anthropic and has a human-like user interface,” Ceceli said. “What I’m saying is that algorithms and data are truly at the core and part of the value creation that we deliver.”
This means there are significant differences in how customers and investors judge AI companies and SaaS startups, and Ceceli says it's important to understand the differences. First, SaaS allows you to bring things out into the world that are far from finished. AI can't do that for a variety of reasons.
“The point is, it's like coding a SaaS product, doing QA, and getting a beta version. It's not a finished product, but you can get it out there and you can start using it,” she said. Ta.
AI is a completely different animal. You can't just put something out there and hope for the best. This is because AI products take time to reach a level where the models are mature enough to work with real customers and be trusted from a business perspective.
“Early on, there is a steep learning and training curve for the algorithm, but the algorithm has to be good enough for customers to want to buy, which means it has to be good enough to create value.” “No,” she said. And this is a difficult line for early-stage startups.
This makes finding early adopters even more difficult. She says she wants to avoid long phone calls where buyers are just trying to learn about her AI. Startup founders don't have time to make those calls. She says it's important to focus on the product and help buyers understand its value proposition, even if it's not quite there already.
“Always be clear about the problem you're solving and the metrics, how you're measuring it,” she said. Optimize for what matters to your buyers. “So you're solving problems around business decisions.” It's okay to be clear about your vision, but always keep in mind your business priorities and how they impact your algorithms. Please use this as the basis of your discussion.
How can AI startups win?
As you build your business, you need to think about how to secure a defensible position with AI. This is especially difficult as major companies continually mine a vast amount of business ideas.
Seseri points out that in the cloud era, there was a foundational layer on which infrastructure players staked their claim. The middle layer where platform players reside. And at the top is the application layer where SaaS resides.
In the cloud, a few players like Amazon, Microsoft, and Google have emerged to control the infrastructure. There are large language models at the foundation layer of AI, with several players emerging such as OpenAI and Anthropic. Although they can claim to be startups, they are not really startups because they are funded by the same large companies that dominate the infrastructure market.
“If you're going to compete for the new foundational tier, the LLM play, you're going to have billions of dollars of capital requirements and it's going to be very difficult, and at the end of the day, it's likely to be a commodity,” she said. I did.
At the top of the stack is the application layer that thousands of SaaS companies have been able to leverage in the cloud era. She said large companies such as Amazon, Google and Microsoft cannot control all of the application layer businesses, and there is room for startups to develop and grow into large, successful businesses.
There is also an intermediate layer where the plumbing takes place. She points to companies like her Snowflake that have managed to build successful businesses in the middle tier by giving application players a place to put their data.
So where is she investing when it comes to AI? “I put money into the application layer and I invest selectively in the middle layer, whether it's proprietary algorithms or open source. , because I think there's a moat around the algorithm, whether it's the data or not. I don't want to own the data. But if I had to choose, I would like to have my own data access and my own algorithms. If forced to choose, I would choose data,” she said.
Building an AI startup is certainly not easy, perhaps even more difficult than a SaaS startup. But the future is there, and companies looking to take on it need to understand what they're up against and build accordingly.