Glasswing's Rudina Seseri says AI has changed the formula for startup success

When Glasswing Ventures' co-founder and managing partner Rudina Seseri appeared at a TechCrunch Early Stage event in Boston in 2024, she already recognized that AI was changing the way firms like hers invest, and what they looked for in a company. The old SaaS playbook was breaking down. You couldn't take an existing product, put a thin AI wrapper around it, add a call to a frontier model, and call it a pivot.

The AI world has come a long way since that appearance, and Seseri's investing philosophy has evolved along with it, but the basics remain the same. "So we have remained extremely disciplined that we don't invest in wrapper companies. We invest in companies that have moats," Seseri told FastForward in an interview last spring.

We've reached a point where coding is the easy part. It's all the rest that goes into building a business that becomes a difference maker. As she sees it, 80% of product development has become generic. It's the remaining 20% in the architecture and the data that will define a company's moat in the future.

This is in line with what Google Cloud's head of startups Darren Mowry told me in an interview last spring. "There have been moments during the hype cycle where people were saying, let's throw the fundamentals out, but there's always a return to, do you have clarity of vision? Do you know who your customers are? Do you know what problem you're solving, and do you have a pathway to profitability? We always come back to that," he said.

Seseri says this changes the way companies operate in a fundamental way. "So the technology stack for sales has changed. The technology stack for developers, of course, has dramatically changed the technology itself, resulting in far greater productivity,” she said. "And the technology stack for marketing, for customer success, it's all a whole new world, and a lot of it is up for grabs, and a lot of it is changing. So it's changing from within, as well as what you are targeting."

Changing investor expectations

The shifting landscape has not only altered how companies are being built, it also has an enormous impact on success metrics and investor expectations. "You know, even two or three years ago, if a company went from zero to $1 million in ARR in its first year in the market, you'd be like, I have a unicorn in the making. Now that number looks more like zero to $2.5 million," Seseri said.

This is directly related to the increase in productivity that AI brings. "So if these guys are doing their job correctly, what they're generating productivity-wise, in code and features, and then customers that are deploying, is orders of magnitude higher, so I'm tying it to the output," she said.

Consider that at HumanX in Amsterdam last week, Max Junestrand, co-founder and CEO at legal AI startup Legora reported $200 million in ARR less than two years after launching.  The company went from $1M to $100 million in just 18 months. It took just six months to double that. That may be an extreme, but it shows that companies can generate revenue faster.

But it's unclear if early revenue is sustainable. Seseri says that these days many customers are trying everything under the sun, so you have to be careful that you're measuring long-term growth and not pie-in-the-sky revenue. 

"I also worry that we're contending with what I call promiscuous customers, who are testing every tool because they're intrigued. So it becomes a leaky bucket. Yes, you generate the revenue, but the retention is not very strong. So there is that phenomenon," she said.

AI's hidden costs

Even though AI clearly speeds up the coding time, and that reduces the amount of people you need early in the history of a startup compared with the past, there's still a high cost to using AI as your programmer, and you have to be careful that it doesn't get out of hand. This is what she calls "the hidden cost of productivity gains," something she says isn't talked about enough.

"So that's not all net productivity. Maybe I should call that gross productivity. The net productivity is the token spend and what's becoming a fixed, variable cost because these are all consumption-based, but it's a fixed line on your P&L of using AI," she said.

As we have seen in recent months, companies are becoming more savvy about looking at costs. The short-lived (and frankly stupid) idea of token-maxxing seems to have outlived its usefulness as companies try to balance usage with productivity and return on investment.

She sees startups and established companies alike experimenting with different pricing models, including outcome-based pricing, but finding the right approach remains a work in progress. "AI will increasingly be about outcomes. So do I charge on an outcome basis? I think all of those are being experimented with, and I don’t think it’s quite clear that it’s a one-size-fits-all.”