What Happens When AI Erases Your Competitive Advantage
AI & Startups
Company Building
Founder Advice
min listen

What Happens When AI Erases Your Competitive Advantage

Published on
November 16, 2026

What Happens When AI Erases Your Competitive Advantage

The Setup

Felipe Ochoa gave himself two years to make IncQuery work. The product took eight months to build, three to four times his estimate, and the first real sale took closer to eighteen. Felipe left Bain in 2013 after running a market research project where version control meant highlighting edits in yellow, then pink, then blue, until the rainbow ran out. He is now CEO of a 90-person bootstrapped company serving consulting and private equity teams, and he has spent the last few years refounding it around AI.

The Risk He Got Backwards

Felipe left Bain believing the market was proven, distribution was a formality, and the product was the open question. He knew the buyers personally and assumed he could sell in ten minutes. The reverse turned out to be true. Friends could not buy a friend's product, vendor certification dragged on, and one procurement form asked what system requirements his browser-based software needed.

The deeper problem was incentives. Consulting is project-based and everyone is working toward the next promotion. The people who care about improving the system sit in central teams with influence but no authority. Felipe describes it as the inverse of bottoms-up SaaS: instead of delighting the user to reach the faraway decision maker, he had to win the faraway decision maker to reach the user.

The Dam Breaks

What finally sold IncQuery was not a pitch. It was one firm going first. Felipe had estimated a good market research project ate 200 expensive consulting hours and bet he could cut that to 40. People believed him in principle. Nobody wanted to be customer number one.

"Once I had the first customer, the dam sort of broke," Felipe says. "They were waiting for a new process. They just didn't want to be the guinea pig." Within months he went from begging for meetings to working 18-hour days, six days a week, on onboarding and database fires. He waited too long to hire because he wanted months of salary in the bank first. His one attempt during the crunch, a warm body off Hacker News, lasted weeks.

The Shutdown and the Hiring Reset

A few months before his wedding, Felipe closed the business for two weeks. No email, no Starlink on the cruise. When he came back he made no announcement and let customers reach out on their own schedule, using the quiet to hire properly. By the time demand returned he had people in place.

His rules from that stretch still hold. Never hire someone you cannot fire, so no friends. Scare candidates with the truth, because self-selection beats your own selection. Being remote before the pandemic made IncQuery a talent magnet rather than a recruiter's client, and he screened for empathy and intellectual humility over pedigree. The first hire was bad. The next ten or twelve were not.

When the Models Showed Up

Felipe's "oh crap" moment came with GPT-2 and DaVinci, well before ChatGPT. IncQuery is a text-heavy business, so the threat was obvious. The collaboration engine that was hard to build in 2013 was no longer the advantage. He treats the company as the incumbent now and runs it on Andy Grove's Only the Paranoid Survive: invest in disruption before the threat is visible, and be willing to cannibalize revenue with a cheaper offering.

Bootstrapping helps here in a way he did not plan for. Without a heavy burn, halving prices is painful but survivable. His new target is another tenfold cut, from 40 hours to four. The moat moved too. "Everybody else's agent has already read Reddit and Amazon," he says. Proprietary primary research is worth more in an AI world, and it is only as good as the questions you put in.

What Worked and What He Would Redo

What worked was the two-year bet, a narrow market he could actually serve, and a customer base that only needed one brave firm. What he would redo is hiring earlier and looking harder for a cofounder, though he would not force one. Felipe also kept one rule from year one that he still credits: no work from midnight Friday to midnight Sunday, because a rule you keep 100 percent of the time is easier than one you keep 99.5 percent. He learned fast. He also learned that the next disruptor should be you.

Tech, startups & the big picture

Subscribe for sharp takes on innovation, markets, and the forces shaping our future.