AI Gives Small Teams Big-Company Output. The Bottleneck Is You

A three-person company can now produce what used to take 30 people. The tools are cheap, the output is real. But the return is just not there. Only 12% of CEOs say AI has delivered both cost and revenue gains, and 56% report no significant financial benefit from it at all, according to PwC's 2026 global CEO survey of 4,454 chief executives. The problem is not access to AI. It is the individual who still has to make every decision and the backlog they create. Chris Poulter founded OSINT Combine, an open-source intelligence software company, in 2019 after more than a decade in the Australian Defence Force, grew it with small teams across more than 25 countries, and combined it with the U.S. firm Kaseware in a 2025 deal backed by the Riverside Company. Poulter says the technology that lets a small team compete can also expose the person at the center of it. 'More technology doesn't make leadership less important,' he says in an interview. 'It makes decisions more consequential because AI amplifies both good and bad judgment.' For a small business owner, that equals a return problem. AI multiplies whatever judgment runs through the owner. When the owner is the only one deciding, AI helps them produce the wrong things faster. The decision is the new constraint For most of the last two years, the small business case for AI was about what AI could do for you. What tasks could AI take on that would make your life easier? And that approach has been working. McKinsey's 2026 State of AI survey of 1,719 respondents found, for example, that roughly a third now skip buying software because a small group builds it internally with AI coding tools. When execution stops being the limit, the issue becomes judgment. The decisions do not get easier when a tool can act in seconds. They happen more often and cost more to get wrong. 'AI can scale execution faster than a founder can scale themselves,' he says. 'Leaders have to distribute judgment, not just tools.' An owner who hands their team five AI agents but keeps every real decision on their own desk has not built capacity. The faster queue still ends at one person. Why the bottleneck now costs more The owner as the sole decision-maker was always a ceiling, and AI lowers that ceiling onto their head at alarming speed. Forbes Daily: Join over 1 million Forbes Daily subscribers and get our best stories, exclusive reporting and essential analysis of the day's news in your inbox every weekday. You're all set! Enjoy the Daily! When output was slow, a founder could review most of it. When a small team ships at the volume of a large one, personal review is no longer possible, and the work either waits for the founder or goes out unchecked. The pressure of a shrinking workforce is looming. AI was the No. 1 reason for job cuts for the last five consecutive months, according to Challenger, Gray & Christmas. So far this year, AI has been cited in 112,713 job cut announcements, approximately 24% of all cuts. Owners are being told AI means fewer people. But the companies that cut headcount without redistributing decisions do not get a leaner company. They are left with the same bottleneck, just now with fewer hands to catch what falls through the cracks. The counterargument: Control is sometimes the edge Distributing judgment early can be the wrong call. In the first stage of a business, the founder's taste is often the product, and handing decisions to a thin team or an unsupervised agent produces drift. I have argued before that a small team is still your real edge over the one-person-company fantasy, and the same logic applies to agents. What makes the difference is what kind of decision is on the table. Reversible, low-stakes, high-frequency calls are the ones to push down and out. The rare decisions that define the brand or bet the company stay with the owner. The failure is treating all decisions as the same and either hoarding them or dumping them. What to do this week Write down the 10 decisions your business makes most often. For each one, mark whether it truly needs you or whether you only think it does because no one spelled out how you decide. The companies seeing measurable AI return are the ones tracking what it earns, not how much of it they run. The same discipline applies to decisions. Move the routine ones out with clear guardrails. Keep the consequential ones. Then watch whether the work still stacks up on your desk. Poulter learned the transition the hard way, building a company that outgrew the founder-with-all-the-answers model. 'As a founder, I had to move from being the person with all the answers to building an organization that could find answers without everything coming back through me,' he says. AI does not make that kind of transition optional. Taking a step back is the difference between a small company that scales and one that stalls at the speed of one person.