Read This Like It Counts

This page is the first step of our hiring process, not a brochure. When you apply, we'll ask what you made of it, and the people interviewing you will have read it too. Read it like something that decides whether you get the job, because it does.

Here's the short version: two years ago we rebuilt this company around AI. The machines do most of the producing now, and the people are here to decide. If that idea gives you friction, this page will help you figure out whether it's the good kind.

What We're Building

Product.ai is founder-owned, profitable, and bootstrapped since 2009. No investors, no board, and more than $1B in annual GMV.

The internet lies to shoppers. Ask an AI what to buy and you get an answer that sounds great and says nothing. We're building the antidote, an intelligence layer for commerce that works out what's actually true instead of averaging opinions. Our patent-pending method, Axiomatic Intelligence, sets rival frontier AI models against each other, and only claims that survive cross-examination make it into our answers.

SimplyCodes is the profitable, growing engine that funds all of this. Around 20 million shoppers a month use it to find codes that actually work. It also shows us what people really buy and return, from millions of transactions the public web never sees. So when the marketing says five stars and our data shows a 40% return rate, we know which one to believe.

We don't sell Search. We sell Closure.

Ask us which camera to buy and you won't get ten blue links. You'll get a verdict, the tradeoffs, and a verified price. As AI agents start doing the shopping, they'll need a source of truth they can trust. We intend to be it.

How We Actually Work

When we say we rebuilt around AI, we don't mean we bought some tools. We mean everyone here directs frontier AI agents that do most of the production work, inside Cortex, the shared written workspace where our strategy, decisions, and shipping live. People and agents work on the same context. It's how about fifteen of us carry the coordination weight of a company many times our size.

AI made execution cheap, so the job that's left is judgment: what to build, what's actually true, what to kill, when something's ready to ship. You're not paid to type. You're paid to decide, and to own what happens next.

It also melted the old job titles. Engineering, product, design, and data have blurred into operators who own outcomes and surfaces instead of task lists. Decisions here don't go to consensus, seniority, or whoever talks loudest. They go to reasoning quality. If you're the newest person in the room with the clearest analysis, you win the argument. Titles don't rescue weak reasoning.

Your outcomes are written down, visible to everyone, and tied to the company's, and feedback comes on a regular cadence. Nobody has to play politics to find out whether their work matters. The system just tells you.

None of this comes free. The bets we've made:

• Stability for speed. Plans change when the facts do. If you need to know what you'll be doing in six months, you'll be anxious here.
• Certainty for autonomy. Nobody hands you a spec or checks your homework. You own outcomes, which means you can also fail in ways that matter.
• Comfort for growth. Your ideas will get stress-tested, sometimes hard. We're trying to find the truth, not win the argument.

We work together, in person, in Los Angeles.

Who Thrives Here

One trait predicts success here better than anything else. We call it architectural comprehension: the ability to hold a large, fast-moving system in your head, see where it's heading, and figure out where you fit without being told.

Pedigree isn't on that list. A brand-name resume is weak evidence either way. Most of the people we want are invisible to normal recruiting. They're inside mid-size companies, at agencies, freelancing, self-taught. If you can show us you do the work, the logo on your resume stops mattering.

The rest is character:

• You own things like a partner, not a renter. You talk about past work as "I built that," not "I was responsible for that."
• You write clearly. Clear writing is clear thinking, and this company thinks in writing. It's the strongest signal we have.
• You hold strong opinions loosely. Being wrong is data. Defending a position you know is wrong is the real failure.

How We Pay

We don't want employees. We want partners, and the pay is built accordingly.

Your salary sits at the 75th percentile, so your life is covered. There's no cash bonus. Instead you get profits interest units in the company with a zero-dollar strike, which means that when their value grows, you pay capital gains around 20% instead of income tax around 40%. Same dollar, and a third more of it stays yours.

You don't wait for an IPO, either. Every December we run a tender where you can sell up to a fifth of your vested units back to the company. It's real money, every year.

If you'd rather have the cash bonus, that's fine, but it means you want to be an employee somewhere, and that's not what we're offering. The exact numbers are in each role listing. We want you to see the math before you ever talk to us.

The Gauntlet

If you're still reading, here's what happens next.

You apply to a specific role at /join/. The written application asks you to respond to this page: what landed, what you'd push back on, and why. Then async video questions built to test how you think, not what you've memorized. Some are deliberately designed so an AI can't answer them for you. What we want is your reasoning, out loud and unpolished. After that come conversations with the founder and the team, then a paid trial: real scoped work at real pay, so both sides find out what it's actually like.

Roughly 1 in 100 applicants gets an offer. The way to be that one isn't a mystery: engage with the specifics of this page, disagree with something and show your work, and bring receipts for things you've actually built. We'll know if you skimmed.

The right person tends to read all this and recognize home in it: how the place thinks, how it pays, what it refuses to do. If that's you, see what's open.