Built to Stay Small: Inside the Org Charts of AI-Native Startups

(Bloomberg) — Amazon.com Inc. spent years hiring its way to dominance — and then hired more people to manage all those people. Chief Executive Officer Andy Jassy is now trying to tame the bureaucracy this created, arguing the company needs to regain its nimbleness and act more like the startup it used to be.

But what if Amazon were founded today? What if its first 10 hires weren’t just “really smart, motivated, inventive, ambitious people,” as Jassy describes them, but AI power users equipped with tools that could handle customer service, scheduling and product prototyping on their own? Could the company have scaled faster, and with fewer employees? 

That’s the underlying premise of a new cohort of startups calling themselves “AI natives” — companies built from the ground up with artificial intelligence not just in the product, but at the heart of workflows and team structure. 

Jared Spataro, Microsoft Corp.’s, chief marketing officer for AI at Work, has been studying tiny, AI-native companies for precisely this reason. “They take a step back and say, what are the inputs to our firm? What are the outputs? And what could we do with AI at the center?” he said in an interview.

While their business models are still largely unproven, founders of AI-native companies say their approach helps them stay surprisingly lean.

At Daydream, a fashion-shopping startup still in beta, the number of staff is in the low dozens. “If you wanted to do this company three or four years ago, you would need so many more people,” said Daydream co-founder Dan Cary, who worked at Google for more than a decade before becoming an entrepreneur.

Or take Lex, an AI-powered writing assistant built for professional writers and academics. It has three employees and its founder wants to keep it that way, even if the business grows. “I love the idea of staying smaller over time so that the core group feels really aligned and really motivated, and able to communicate about hard things,” said founder Nathan Baschez. “You want the right small group of people rather than more people.”

It’s not just the raw computing power of AI that keeps headcounts low — it’s AI’s ability to collapse work that normally spans multiple functions into tasks that can be accomplished by one person, if any.

At Daydream, even product managers who aren’t highly technical can use AI coding assistants to build prototypes themselves, rather than having to wait on designers or engineers to build something for them. “It removes the technical barriers to creation” and speeds up the work, Cary said.

Rather than testing one idea at a time, like a traditional company might, Daydream typically tests 15 to 20 ideas in parallel. “These AI tools are phenomenal at helping you find out if you’re wrong faster, and helping you get to right faster,” he said.

That’s getting easier with the rise of AI agents, which are capable of handling multistep tasks. While a generative AI chatbot might be able to craft an email to a customer, an agent can be built to actually send the email, log any replies and even schedule a service call if needed, for example.

At Sailplane, a stealth-mode startup building agents for AI engineers, co-founder and CEO Sam Ramji’s scheduling is handled entirely by an AI agent that mimics the tone and cadence of a professional assistant. Ramji said the company plans to build agents for any parts of the business where off-the-shelf AI solutions aren’t yet available.

Sailplane has a team of six and aims to close out the year with no more than 10 people.

With small headcounts, founders say they can devote more time to building and less time to traditional management. Lex’s Baschez estimated he spends 70% of his time on engineering work, a luxury he suspects he would lose if the business added many more people.

He thinks large companies are disproportionately harder to manage and likens them to animals raised on factory farms. “They grow them way faster than they’re supposed to and it causes all sorts of problems,” he said.

AI-native companies have a chance to keep their org charts smaller and also more fluid, even compared with other startups. While it’s not unusual for employees of early-stage companies to wear many hats, one of the promises of AI is that it will extend the range of each person, like the product managers now making prototypes at Daydream.

Microsoft’s Spataro said he’s noticed that AI-native firms tend to employ more generalists. Since AI tools can often handle narrowly defined tasks traditionally given to specialists​, “understanding the business end to end proves more valuable,” he said.

But working at an AI-native startup still requires at least one special skill: proficiency using AI. 

Shortwave, a startup making AI-powered tools for managing email, cut its staff to six from 15 after parting ways with employees whose AI use wasn’t up to par, said founder and CEO Andrew Lee. While he plans to add back a few more people, keeping a small headcount doesn’t worry him. He’s betting on swift execution to stay competitive — powered by agents, not layers of middle management.

“We need to optimize our team for a small number of highly cross-functional people where we can move super fast,” he said. “Our moat is going to become speed.”

More stories like this are available on bloomberg.com