An accountant supplied Ahmad Ibrahim with a useful correction to the vocabulary of Silicon Valley. Ibrahim was working on QuickBooks Accountant at Intuit when a colleague asked a customer whether a feature was delightful. The response, as Ibrahim remembers it, was brisk: “At best, you’ll make it less painful.” A small sentence, with rather large implications for anyone trying to sell accounting software.
“Delightful” is a lovely word for a pastry. It becomes a more demanding proposition when attached to a tax workflow. The customer had supplied a different measure of success: remove enough unpleasant work that someone notices its absence. Ibrahim’s subsequent career has kept returning to that measure. How much effort does a product demand before it starts giving something back?
He is the co-founder and CEO of Neo.Tax, based in Mountain View, California. The company automates research-and-development tax-credit work and software capitalization. His part in its founding team comes from product management; his collaborators brought machine-learning research and tax expertise. The combination puts an unusually practical question beneath the AI discussion: can a company account for its work without asking everyone to describe that work again?
A useful idea meets an uncooperative customer
Ibrahim studied economics and philosophy, attending UC Berkeley and the London School of Economics. His first job was in structured finance at Merrill Lynch. An earlier venture, Unleash, attempted to give small businesses CFO-style insight through an app. It depended on accounting records that customers often had little interest in keeping current. An elegant model had encountered an inconvenient human habit.
There is a particular frustration in building software that would work beautifully if its users first became different people. The product can be internally coherent and commercially stranded. Its inputs may exist in theory, its recommendations may be sensible, and the customer may still have better things to do on a Tuesday afternoon. A request to tidy the books is also a request for time.
Ibrahim’s account of that venture is candid about the limits of persistence. He had been trying to persuade customers to recognize a problem rather than responding to a problem they already wanted solved. That distinction helps explain his attraction to tax: the obligation arrives whether a business owner feels enthusiastic about it or not. Demand does not require a motivational speech.
Philosophy supplies a useful background for this sort of reconsideration. An assumption can look sturdy until someone asks what must be true for it to hold. In a product, that question eventually becomes concrete: where will the data come from, who will maintain it, and why would they bother? The answer has to survive contact with a customer’s calendar.
Ten hours a week, then a company
The early company was called cpa.ai. Ibrahim approached Firas Abuzaid, whose Stanford graduate work covered machine learning and systems, about automating the R&D credit. Abuzaid initially declined. They stayed in touch. When Ibrahim returned with a working early product and customers, the invitation became smaller: would he contribute ten hours a week?
Those ten hours grew to twenty, then forty, in Abuzaid’s recollection. The progression is a better origin story than a polished scene in which three founders simultaneously discover their destiny. Someone made an invitation. Someone said no. Work continued. Another invitation contained something more persuasive than an idea: evidence that the idea had begun to move.
Stephen Yarbrough, the third co-founder, brought experience auditing R&D credits at the IRS. The founding division of labor was explicit: Abuzaid’s technical expertise, Yarbrough’s tax knowledge, and Ibrahim’s product judgment. Each discipline has its own definition of a satisfactory answer. A system needs to meet all three before a customer can use it comfortably.
Neo.Tax’s stated values include mastery of the domain, transparency, and simplifying complexity. These sound easy enough until the disciplines disagree. A tax specialist may want more detail; a product designer may want fewer steps. The work lies in preserving the detail while reducing the effort required to understand it. That is a substantial part of the promise Ibrahim has chosen to sell.
A bank customer becomes a bank partner
In February 2022, Neo.Tax announced a $10 million Series A led by Infinity Ventures, with participation from GV, Acrew Capital, Fin Venture Capital, and existing investors. The announcement also brought a Mercury partnership. There was a pleasingly ordinary beginning to that relationship: Neo.Tax had itself been a Mercury customer.
Ibrahim emphasized shared values when discussing the partnership. Mercury’s business served startups; Neo.Tax wanted to help those businesses access R&D credits. A banking relationship became a route to customers who might otherwise have had to discover another specialist product on their own. Distribution had become part of the product’s usefulness.
The distinction matters in Ibrahim’s story. A well-designed service still needs to arrive where a customer can find it, at a moment when that customer can use it. The founder who understands the workflow also has to understand the introduction. Money in the bank pays for development; a partnership can put the resulting work in front of someone with an actual need.
In 2023, Neo.Tax announced a CFO advisory council. Ibrahim described the purpose as working closely with finance leaders to understand their needs. The company also received a 2023 Product Award in the FinTech category from Products That Count. Recognition and advice serve different functions: one acknowledges a product, while the other gives its makers more opportunities to hear what needs changing.
Led by Infinity Ventures
The filing needs a trail
The appeal of tax automation becomes clearer when Ibrahim explains where the software should look. He argues that it should examine the business’s records and put them in the context of tax rules. A tax code contains the instructions. The company contains the facts. Bringing the two together is the labor customers are trying to reduce.
In his April 2025 writing about R&D credits, he broke the work into three problems: identifying qualifying work, determining the time spent on it, and producing technical explanations suited to a tax filing. He argued for using existing project and payroll data rather than repeatedly collecting people’s retrospective accounts. His case is grounded in the documentation a business already creates.
Consider the burden implicit in an interview about work completed months earlier. Somebody has to remember. Somebody has to ask follow-up questions. Somebody has to turn the answers into a usable explanation. Each handoff costs attention. Ibrahim’s preferred direction is to make the original records useful to another department without turning the engineers into part-time narrators of their own working lives.
Trust remains central to the pitch. In 2023, Ibrahim discussed the need for AI software to explain its conclusions and acknowledge low confidence. “You can’t hallucinate in taxes,” he said. A finance team needs to understand the path from an input to an answer. An attractive interface cannot excuse an answer that nobody can account for.
That makes explanation a product requirement. Speed has little practical value if a reviewer must spend the saved hours figuring out whether the result is defensible. The customer needs a way back through the work: the record, the treatment, the calculation. Ibrahim’s approach treats that route as something the software must help preserve.
- Read existing recordsProject activity + payroll data
- Apply tax criteriaConnect the work to the rules
- Explain the calculationPreserve a route back to the evidence
The product person learns to sell
By June 2026, Ibrahim was describing enterprise tax departments as Neo.Tax’s audience. He named Adobe, Block, Capital One, and Databricks among the companies using it. Those are his reported customer examples. The scale of the intended customer had changed from the early small-business proposition to organizations with substantial tax operations.
His account of his own job had changed too. He had once thought of himself primarily as a product person, with commercial work somewhat outside that identity. He now puts selling near the center of the CEO’s responsibilities. “Every CEO has to sell,” he said. The lesson concerns what selling teaches the founder as much as the revenue it brings in.
A sales conversation is a demanding kind of product research. A prospect has reasons to hesitate, people to consult, and a way of working that has survived previous software pitches. The founder has to explain what will change, listen to what cannot change, and learn which benefits matter enough to justify the disruption. The conversation can be uncomfortable precisely because it is useful.
The enterprise move also involves people who already know that world. Jo Bryant joined Neo.Tax after more than two decades at Thomson Reuters. She described Ibrahim introducing her to the product and team after the companies’ reseller agreement. Her account emphasizes both the difficulty of winning over skeptical tax departments and the collaborative culture she found inside Neo.Tax.
Ibrahim’s public posts keep returning to the need for repeatable, auditable outputs. He describes Neo.Tax as an engineered system, with many model calls and internal structure, rather than something a team can reproduce by finding a clever prompt. It is a revealing emphasis for a former product manager: the experience looks simple only after considerable work has made it reliable.
“Every CEO has to sell.”Ahmad Ibrahim · June 2026
A bigger ambition, with the same obligation
Ibrahim’s ambition extends beyond the R&D credit. He says customers have asked Neo.Tax to automate more of their taxes, and he frames that broader automation as the company’s destination. Its current mission includes R&D credits and software capitalization, with further expansion in view. The starting problem remains specific even when the intended business is much larger.
He also takes the work beyond private sales calls. He has appeared in tax-industry webinars, including a February 2024 event with the Tax Executives Institute and Thomson Reuters. In 2025, TEI again listed him as a speaker on R&D credits and reporting changes. These are audiences for whom a persuasive demonstration must withstand technical questions, not merely earn applause.
The trajectory has a certain consistency. The earlier app asked customers for better inputs. The later company works to extract usefulness from records already being produced. The product manager learned that removing an annoyance was a serious achievement. The CEO learned that he needed to sell that achievement himself. Each adjustment brought the idea closer to the person expected to pay for it.
Taxes supply few opportunities for theatrical joy. A finished filing rarely gets framed on the wall. That leaves Ibrahim with a quieter standard: the work is supported, the customer understands it, and fewer people have spent their afternoon reconstructing the past. The accountant’s correction still holds up. Less painful is a perfectly respectable ambition.