How will AI reshape company workforces? They'll be a lot smaller and flatter, if startups like Pointhound are any indication.
The three-year-old company, which helps consumers book flight deals with credit-card points, is the second business Jay Reno has launched in a decade. The first, a subscription-based furniture rental company he started in 2017, employed around 150 staff at its peak before he sold it in 2022.
Pointhound, on the other hand, has four full-time human staffers and a fleet of artificial-intelligence agents, even though 750,000 people used its products last year. Projects that took six months at his old company now take days, says Reno, Pointhound's chief executive and co-founder.
It is like "everyone was just given the ability to take a GLP-1," says Reno, comparing the AI's slimming effects on companies to weight-loss drugs. Even if Pointhound's sales grew 10-fold overnight, he estimates he would maybe need to hire just one more engineer and someone to help with marketing.
The newest generation of companies, infused with AI from the start, offer a vision of how work could soon be structured elsewhere in American corporations: Fewer co-workers; more on-staff engineers; and a flatter structure in which nearly everyone is a player-coach instead of strictly overseeing teams.
A new working paper examining thousands of recent Y Combinator and other U.S. venture-backed startups indicates as much. The study found that AI-powered startups have about 25% fewer employees than comparable companies not built around AI capabilities.
They had about 15% fewer entry-level workers, 15% fewer managers, and fewer rungs in the corporate ladder -- yet with similar valuations as their less AI-focused peers.
In short, they are the proving ground for the philosophy that corporate giants like Amazon.com and Meta Platforms are now channeling as they reorganize, de-layer, cut jobs and invest in more AI.
When Microsoft's Xbox division said in early July that it would lay off 3,200 people, for instance, it pledged to cut management layers to between three and five. Before, work in some parts of the company went through as many as 14 layers, and Xbox's platform teams have swelled even as its player base has declined, its CEO said.
"That complexity has slowed decisions, blurred accountability and made it harder to deliver," Xbox CEO Asha Sharma wrote to staff. "We will simplify."
Startups have long been harbingers of broader workforce and technology trends, such as the rise of the product manager and the shift to "digital-first" business models.
Today, these AI-native companies -- built from scratch with AI in their DNA -- serve as the first trial of the management playbook for the agentic era.
"Paying attention to startups is just so valuable right now -- they're going to tell you where things are going," said Rem Koning, an entrepreneurship professor at Harvard Business School who led the startup research with Insead professor Hyunjin Kim.
Employees at the AI-native startups were more likely to come from prestigious companies and hold degrees from elite schools. They were also more likely to be male, the researchers found. The companies raised similar amounts of money, so the AI startups had about 20% more capital per employee.
The research found startups that provided services -- for example, in mental health and tutoring -- had the most drastic drop-off in staffing. Those AI companies were 70% smaller than non-AI companies.
Slimmer companies don't necessarily mean fewer people with jobs, Koning said. AI makes it easier to create businesses, too, and so may result in more companies, albeit with smaller workforces.
He believes established companies won't achieve significantly more productivity just by giving AI tools to employees. The streamlined reporting lines and management layers make a big difference, as does embedding AI into work flows.
"There's going to be a hard limit on what existing organizations are going to be able to achieve," said Ori Eldarov, the CEO of AI-powered investment bank OffDeal, which raised $12 million in Series A funding last summer. Many companies are organized around human-to-human interactions, not human-to-AI interactions, he points out. "Baby steps," he said, avoid risk but won't fix inefficient workflows.
At OffDeal, engineers and bankers sit side by side, and pay for both groups is tied to closed deals. With nearly a 1:1 ratio of engineers to bankers, it's a far cry from how many Wall Street firms position engineers, who are often holed up in New Jersey back offices, Eldarov said.
The company had software engineers shadow bankers for two weeks from 8:30 a.m. to midnight and catalog everything they did to find efficiencies.
The exercise revealed that pre-call prep and post-call debriefings could be automated and that many reports could be scrapped. Engineers also built an AI-powered daily task-management system, which bankers wouldn't have known to even ask for, Eldarov said. Before, it had been the bankers' jobs to find prospective buyers for client businesses, heavily assisted by automated tools. But after the shadowing process, he reassigned that task to its business-development team.
Much of that work is typically handled by entry-level analysts at other firms, Eldarov adds. But "had we hired junior bankers day one, we would not have been forced to innovate on the process."
Write to Lindsay Ellis at lindsay.ellis@wsj.com
(END) Dow Jones Newswires
July 19, 2026 20:00 ET (00:00 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments