5x output, not 5x revenue: three founders on what AI really changes
A Los Angeles product chief, a Brisbane founder and a Kharkiv CEO on what happens after AI hands them the hours.
A Los Angeles product chief says AI lets each employee do roughly five people's work. Then comes the catch: "Revenue is nowhere near five times higher."
Productivity went up. The bottleneck just moved.
Two founders, in Brisbane and Kharkiv, report the speed-up from the other side: real hours won back, and people still in the seats.
Here is what each of them kept for humans, and what that means for anyone weighing a hire against an agent.
We asked which parts of a job AI has actually taken, and which still need a person. The answers came back in hours, percentages and one very stubborn bottleneck. It matters now because every company is deciding what to hand to an agent, and the hours are only half the math.
The searching went to the agent. The team stayed.
Andrew Priobrazhenskyi, Founder and CEO of DiscountReactor (Kharkiv, Ukraine)
“At DiscountReactor, the searching moved to AI completely. My agent checks about 2,000 retailers every day, work that used to take us 15 hours a week by hand.”
“With a team of four, the agent took over about 60% of our research time, and nobody lost their job.”
A clean handoff: one repetitive task moved over entirely, and the people got their time back. Notice what did not change: the headcount.
Ten years of judgment, now with a faster engine.
Roo Wright, Founder of SignalTo.ai (Brisbane, Australia)
“One of my developers has worked with me for nearly ten years. He's gifted, but speed was always a trade-off. Since we introduced Claude Code and learnt how to use it properly, I've seen his productivity increase by at least 150%.”
“He knows when the AI has misunderstood the job or suggested something that will cause problems elsewhere.”
The speed-up is huge. The veteran is still the one who catches the bad suggestion, and that part stays human.
Five times the work. Not five times the revenue.
Tejas Choudhary, Chief Product Alchemist of Outrider (Los Angeles, United States)
“At Outrider, AI lets each employee do roughly five people's work. Revenue is nowhere near five times higher. Decision makers, clients and government bureaucracy are still the bottlenecks.”
Doing the work faster is one problem. Getting the people around you to move faster is another, and that is where the revenue gets stuck.
Where they split
Roo Wright and Andrew Priobrazhenskyi report real time won back: at least 150% more productivity for one developer, and about 60% of research time for a four-person team.
Tejas Choudhary asks what that time is worth. In his telling, decision makers, clients and government bureaucracy set the pace, not the tools.
None of the three describes taking people out of the loop. They differ on where the gains show up: in the workday, or in the business.
My take
AI speed is a fact. Business speed is a separate question. Before you count the hours saved, find the bottleneck those hours were never going to fix. Then sort the work: what can the agent own outright, and what needs a person with something to lose if it goes wrong. If the honest answer is only time saved, enjoy it, then keep looking for the revenue.
Your turn
Where did AI give you hours back, and what stopped those hours from turning into revenue?
Next: an Australian founder lets AI read and filter every press request, while an Estonian one-person store and a US PR tool keep any message to a journalist behind a human approval.
