I feel the need for speed

In the 1986 movie Top Gun, Tom Cruise’s character Maverick and his pal Goose famously uttered the line, “I feel the need…for speed.” It’s applicable in so many contexts, but never more so than for startups racing to keep up with the changes AI is bringing to technology. 

It’s conference season, and I attended a couple of events recently that brought that idea of speed to the top of my mind. One story in particular stood out, a startup that was recently acquired after quickly making a bold AI pivot.

That startup is Neon, a Postgres database company that Databricks acquired in May for $1 billion — a nice exit for a four-year-old startup that had raised close to $130 million, per Crunchbase data. 

Speaking recently at the Madrona IA Summit, Neon co-founder Nikita Shamgunov told the story of his company’s rapid shift from a database geared toward human developers to one designed for autonomous agents — and how that shift changed the company’s entire trajectory. It was a story of velocity and the ability to adapt quickly, try new approaches and toss aside what didn’t work. It’s a lesson that many startups can learn from. 

Neon’s billion dollar pivot

Shamgunov explained that early on, the company was building a serverless, Postgres database for developers. Seemed like a reasonable target market, but the company would soon learn it wasn’t the correct one.

For obvious reasons, the company’s initial focus centered around building the best developer experience possible, and the approach seemed to be working. “We were very happy because there were a bunch of nerds twisting knobs on the back end with more usage and better retention,” he said. As long as you had happy customers, and the numbers were going up, why would you change?

Then something serendipitous happened that would help future-proof the startup, something that would require them to recognize an opportunity when it presented itself. The company decided to embed the database in developer platforms, hoping for more uptake. One of those platforms was Replit, which would itself soon be transformed by AI. But at the start of the relationship they weren’t seeing much additional usage, that is until Replit released its Agent product last fall.

With Replit Agent, developers could create code from a prompt and that produced an unexpected phenomenon for Neon. The company began to see more databases being created by Replit than in any other channel, and realized that the agents were rapidly generating their databases in much greater numbers than human developers. While that played well to the serverless aspect of the product, making it easier to spin up the databases as demand required, Neon also realized that this shift would require rethinking the product in this new agentic context. If they had failed to see Replit as the edge of an AI-driven sea change, they quite possibly could have been left behind. It was another lesson for founders: pay close attention to your usage patterns.

Rapidly shifting gears

Speed was of the essence, and as Madrona’s Jon Turow, an investor in Neon put it, the company began “a ferocious pace of execution.” What Neon did was not only recognize the shift, but react to it swiftly. Shamgunov said they started a bunch of simultaneous experiments. While the company didn’t get into specifics, it was more about the process they undertook to successfully transform while taking calculated risks.

“When you lean into something that's new and uncertain, you’ll take a lot of shots and inevitably miss some,” he said. That’s far better, he argued, than being so cautious you only take a handful of perfect shots, or worse, never act at all and miss the bigger opportunity.

They made a choice to embrace the disruption, to look at the shift to agentic AI and what that meant for the product and the market, and it worked out well for them. Databricks saw what they were doing and acquired them in May for $1 billion. Today Shamgunov is a vice president at Databricks, a company valued at $100 billion, following its Series K financing in August, with an IPO very likely in its future.

If you’re a 4-year-old startup competing in a crowded market like Postgres databases, if you see a chance, and it pays off like this, I think most founders would happily take this outcome.

As Shamgunov says, every company has the same choice when a major change like AI comes along. You can either embrace it and the market perceives you as a modern company, or you deny it and get left behind. It takes courage to make a massive change in direction, but it also takes that “ferocious ability to execute.” Neon made the right choice, accelerating when the circumstances demanded it and getting rewarded handsomely for it.

~Ron

Featured image by Srikanta H. U on Unsplash