Build AI that is safe and accessible for the world

We’re a for-profit company. Of course profits matter. But profits aren’t our focus. We build amazing products. We build healthy margins into those products. So profits aren’t something we’ll ever have to chase.

What we care about most are the people who trusted us when we hired them. And we care about the customers who have trusted us to use and purchase our products. And we care that the AI we make accessible is safe for our customers and for the world.

We believe that to hire the best, you have to go where they are. You don’t make them move across the country. You don’t make them commute two hours each day. You don’t threaten them and mandate days in the office or track them with swipes and software.

Hiring the best also involves trusting who you hire. When you allow people to work how they want to work and trust that they will deliver, we believe that is when you build a sustainable culture. Because culture isn’t ping pong tables and a snack bar (not that we’re anti-snack; we’re very pro-snack here).

We start with the person using what we build. Not with a competitor’s launch, not with what’s fun to build and not with what’s easiest to ship. We read what customers tell us, we watch how they actually use our products, and we fix what gets in their way, even when nobody asked us to. When someone trusts us with their work, the least we can do is sweat the details for them.

We never war with one another. Disagreements happen. But we don’t war. We strive to build a team with low attrition, but if you war, you will quickly find the exit door.

Not all revenue is created equal. We will never engage with building AI systems for war, apartheid, ethnic cleansing, terrorism, illicit activities or racial profiling or intimidation. We respect and encourage our team to develop new ideas across the spectrum of verticals, but never for areas that disagree with our ethics.

If you see it, you own it. That goes for your own work and for the bug you tripped over in somebody else’s. Owners don’t wait for permission to care, and they don’t hand off a problem half solved. They think about how a decision will look in a year, not just at the end of the week. And when something goes wrong, owners are the first to say so.

The best answer usually has fewer parts. We’d rather delete a feature than explain it, and we’d rather ship something simple that works than something clever that needs a manual. Good ideas come from anywhere, including the person who joined last week, and we’d rather try one and be wrong than never try at all.

AI moves fast enough that what you knew last quarter might already be out of date. So we read, we tinker, we ask the questions that sound obvious, and we change our minds when the evidence tells us to. Nobody here is expected to know everything. Everybody here is expected to want to find out.

Good enough usually isn’t. We hold a high bar for our products because our customers and their data deserve it, and we hold a high bar for each other because that’s how we all get better. When something falls short, we fix the cause instead of patching around it. And we don’t ship anything we wouldn’t be proud to use ourselves.

We’re building Boo, an AI agent that runs open and closed-weight models. We’re also building the YottaBot Agentic Operating Layer, which fleets of agents run on. Neither of those is a small idea. We think about where things could be in five years, not just next week, and then we break that down into steps we can actually ship.

Most choices are easy to change later. For those, we’d rather decide quickly, see what happens and adjust than wait around for a perfect answer that never shows up. The few that are hard to undo get the time they deserve. Knowing which is which is part of the job.

Trust gets built in small moments: saying what you’ll do and then doing it, owning it when you get something wrong and giving credit where it’s due. We’re direct and we’re kind, and we don’t think those are opposites. Our customers trust us with their data and their work, and we never take that for granted.

If you disagree, say so, respectfully and with your reasons. Nobody gets points for nodding along in a meeting and grumbling about it afterwards. Once we’ve decided, we all get behind the decision, including the people who argued against it. And if something new comes to light, we’re happy to open it back up.

Ideas are great. Shipped ideas are better. We focus on what actually matters to our customers, we do it well, and we follow through when it gets hard. Remote doesn’t mean invisible: your results speak for you, wherever you happen to be sitting.