Ayush Sharma

Somewhere in a state department portal, there is a form. It asks for an account number you do not have, references a rule nobody explained, and threatens a penalty if you miss a deadline you were never told about. Ayush Sharma has met that form in dozens of states. He found it so insulting to the act of building a company that he started one to kill it.

That company is Warp - an AI-native employee management, payroll, and compliance platform built for high-growth startups. The pitch is unglamorous and exactly the point: register your company in every state, run payroll for W2 employees across all 50 of them, pay 1099 contractors anywhere, and settle international workers in 150+ countries, without the founder ever touching a government website. Today Warp does this for more than 380 companies out of a Flatiron office in New York, and in June 2025 it raised an $18 million Series A led by Sound Ventures.

Sharma's framing for what he is building is blunt. "We're building the AI-native infrastructure startups actually need," he says, "tools that think, decide, and act." That last verb is the whole company. Plenty of software shows you a checklist of compliance tasks. Warp wants the software to do the tasks - file the registration, fetch the certificate of good standing, watch the deadline, handle the notice - so the human can go back to the product.

It is a strangely personal mission for something as dry as unemployment insurance filings. But that is the tell. Sharma did not arrive at payroll because it was a hot market. He arrived because he had personally been buried by it.

Before Warp, Sharma was deep in code. He studied both computer science and physics at MIT, then stayed for a master's in computer science focused on systems and machine learning. He went on to engineer at Yelp, where he worked on core ML infrastructure and improved the models that classify photos - the unglamorous machinery that decides whether the picture you uploaded is a burger or a storefront. Earlier still, he worked in product and growth at Rize Education and did undergraduate research at MIT.

The lesson that became Warp did not come from a lecture hall. It came from his previous startup, where he hired and ran a fully distributed team spread across multiple US states and several countries. Two things became obvious fast. The first: distributed work is not a fad. "The future of startups is distributed," he wrote at launch. "After COVID, in 2023, we're still only at the beginning of this revolution." The second was the catch nobody warns you about. Every state you hire in comes with roughly three separate agencies to register with, each with its own portal, its own jargon, its own clock.

So when he and a small team of engineers launched Warp out of Y Combinator's Winter 2023 batch, the product was not a guess about a market. It was the tool he wished he had owned the first time around. "With Warp, I'm building the product I wish I had in my previous startup," he says. Four weeks after starting, they were already running real payroll for YC and non-YC startups alike.

The incumbents, in Sharma's reading, made a quiet bait-and-switch. You sign up for payroll, assume the job is done, and then the emails start - state tax registrations, foreign qualifications, unemployment insurance, ongoing reports, franchise taxes. The software told you what to do and left you to do it. Meanwhile, the tools had ossified into bloated suites built for HR departments, gated behind sales calls and upsells.

Warp inverts the deal. Onboarding takes about ten minutes. From there the platform registers with state tax agencies, obtains certificates of good standing, helps with foreign qualifications, and monitors deadlines on the company's behalf. The numbers Warp reports for the approach are the kind that make the pitch land: filing time cut roughly 10x, error rates down about 80%, and cost savings north of 90% against legacy providers. The detail that says the most - Warp claims it handles this volume of compliance with essentially one part-time contractor. The rest is the machine doing the filing.

In June 2025, Warp announced an $18 million Series A led by Sound Ventures, the firm co-founded by Ashton Kutcher. That round pushed total funding to roughly $24-25 million. The institutional list is solid on its own - Y Combinator, HOF Capital, SV Angel, Homebrew Capital, plus A-Star and Abstract. The angel list is where it gets fun.

Backing Sharma personally: Dropbox co-founders Drew Houston and Arash Ferdowsi, Cruise founder Kyle Vogt, Replit's Amjad Masad, and Balaji Srinivasan - alongside operators from OpenAI and Brex. It is a roster of people who have each built the kind of company Warp is selling to. They are not buying a payroll vendor. They are buying a bet that back-office work belongs to agents now.

Plenty of companies bolted the word "AI" onto an existing product in 2024 and 2025. Sharma's claim is sharper, and it shows up in how he describes the work. Warp did not start as a payroll company that later added intelligence. It started as a company whose core operation - reading obscure government rules and acting on them - happens to be exactly the kind of repetitive, language-heavy, rule-bound labor that machines are now good at. Compliance is not a feature to be automated later. It is the thing the software was built to do from the first line.

That reframing changes what the company is selling. A traditional payroll provider sells a system of record: a place to store who works for you and how much they make. Warp, in Sharma's telling, sells action. The product registers the entity, files the return, fetches the certificate, and clears the notice. The human's job shrinks to approval and exception-handling. When he says he wants tools that "think, decide, and act," the order of the verbs is deliberate - most software stops at "show." His whole argument is that the last two verbs are where the value now lives.

There is a quieter thesis underneath, too. Sharma believes the shape of the modern startup has permanently changed. A company three people strong can now have employees in four states and contractors on three continents before it has a single paying customer. That is wonderful for talent and terrible for paperwork, because the regulatory map did not get any simpler - it got multiplied. Warp is a bet that the only sane response to that multiplication is to stop asking founders to keep up by hand. The market he is chasing is not "payroll." It is every minute a founder loses to the machinery of merely existing as a company.

Sharma talks through his view on AI in product development in this conversation: NachoTips: AI in Product Development - Warp CEO Ayush Sharma.