He looked at the most dreaded chore in business - filing taxes - and decided a machine should do it. Not better. Gone.
An accountant once leaned across a product demo at Intuit and told Ahmad Ibrahim the truth nobody at the company wanted to hear: "You're never going to make it delightful. At best, you'll make it less painful." Most product managers would have filed that under feedback and moved on. Ibrahim heard a thesis statement.
He runs Neo.Tax, the Mountain View startup he co-founded in 2019 to automate tax filing - beginning with the R&D tax credit, the gnarly, paperwork-heavy benefit that puts real money back into companies that build things. The company has raised roughly $13 million to chase a deceptively simple idea: taxes are lots of rules and lots of numbers, which is exactly the terrain where computers shine and humans wilt.
The trick, in his telling, is knowing where to aim the machine. "You can't hallucinate in taxes," Ibrahim says, and he means it as a design constraint. Point an AI directly at the tax code and you get a confident, occasionally wrong intern. Point it at a company's actual transactions - and let it map them onto the rules the way a careful accountant would, line by line - and you get something that can produce audit-ready filings instead of guesses.
Neo.Tax's product turns the R&D credit, including multi-year lookback claims, into output a company can actually defend, without the surveys, interviews and back-and-forth that normally make the process miserable. The pitch is less "trust the robot" and more "the robot reviewed every single document, and it will tell you when it isn't sure."
"No one enjoys doing this shit."- The realization at Intuit that became a tax-AI company
Ibrahim studied economics and philosophy at UC Berkeley, then went into structured finance at Merrill Lynch. He is the first to admit philosophy "didn't help me get my first job." It paid off later, when the job became evaluating markets, stress-testing assumptions and deciding which obvious-looking ideas were actually true.
Then came a detour most founders don't have on the back of their business cards. From 2015 to 2018 he was co-founder, COO and CFO of Port of Mokha, a specialty coffee company sourcing beans from Yemen. Operations and finance for a physical, supply-chain-heavy, geopolitically complicated product is a strange apprenticeship for a software CEO. It is also a very good one.
His first software startup, Unleash, tried to deliver CFO-grade insight by reading a small business's QuickBooks data and handing back benchmarks and forecasts. It died on a single human fact: small business owners would not keep their books updated. "Bookkeeping is the worst part of running my business," they told him. The lesson stuck. "When you're convincing your customer why they have a problem that they're not even aware they have, you're doomed."
At Intuit he built QuickBooks Accountant from zero to one - close enough to the pain to feel it daily. Add it up and Neo.Tax isn't a pivot. It's the conclusion of an argument he'd been assembling for a decade: the rote, computational core of accounting is exactly the kind of work software should swallow whole.
Pointing AI at the raw tax code, he argues, is the seductive wrong answer. Aim it at real transactions and let it contextualize them inside the rules - the way a working accountant does.
"The success of your AI product is going to come down to - people need to trust it." The tool that admits when it isn't confident beats the one that's confidently wrong.
Much of accounting gets automated away, he believes. The best accountants won't be replaced - they'll be the ones wielding the AI, freed for the creative, advisory work.
"Customers get the best experience when the team is at their best." He treats team excellence as the primary lever for everything else - including the bottom line.
"You can't hallucinate in taxes."
On why accuracy is non-negotiable"No one enjoys doing this shit."
The Neo.Tax origin, in five words"90+% of your waking hours go into your startup. It might as well be around great people you enjoy."
On why he hires for people first"When you're convincing your customer they have a problem they're not aware of, you're doomed."
The lesson from UnleashHe ran finance and operations for a Yemeni specialty coffee company before he ever shipped tax software. Supply chains to the IRS - a plausible career, apparently.
The hardest founding hire wasn't the Stanford ML PhD. It was finding the right former IRS agent who was forward-thinking and didn't feel threatened by automation.
A job candidate once told him, after a team lunch: "I just listened to you guys talk and everyone likes each other. I don't think I've seen that before."
He says philosophy was useless for landing his first finance job - and indispensable for everything after it, especially stress-testing which "obvious" ideas are real.