"If we don't raise $10 million with this team and this idea at this time…"

That was the bet. We had no deck. We were betting on this team and a conviction we had spent six months and our own money arriving at: the next great companies would be built by humans and AI coworkers operating as one team.

We gave ourselves ten days.

First investor call: March 12. Term sheet: March 26. Fifteen calendar days.

The market now has a name for part of that conviction: multiplayer AI. Humans and agents share context, work, and responsibility instead of one person disappearing into a private chat window with an assistant. The name is new. Our bet went further: a team can move at forty times its old speed only when the humans know how the others arrived at what they believe.

AI makes execution abundant. It does not make a team believe the same outrageous thing at the same time.

Trust does that.

Forty times faster is mostly less waiting

When people hear that SageOx operates at 40x, they assume the multiplier comes from AI coworkers writing code quickly. That is half of it.

The other half is the time we don't spend waiting: for the polished version, for somebody to admit what failed, for a decision made in one head to reach every other head, for the private conversations after the meeting, for the founder's mood to become safe enough for the true thing.

AI coworkers make the work move faster. Radical transparency makes the truth move faster. At 40x, the bottleneck is not how fast you can do the work. It is how long the truth spends in somebody's pocket.

In a spinning mill sometime before 1942, one operator's thread almost never broke. Every other operator was constantly fighting theirs. The supervisors watched him and could not see what he did differently, so they finally asked him to name his price for the secret: a pint of beer every day for the rest of his life. The bargain was struck. He gave them three words: chalk your bobbins. He kept a piece of chalk in his pocket and rubbed its dust onto each wooden bobbin so the thread would not catch on fine splinters. The solution had been twenty feet away for years, trapped in one person's pocket, while the rest of the mill paid for not knowing it. The story was recorded in Earle Buckingham's 1942 Production Engineering. I found it in Brian Potter's The Origins of Efficiency.

AI today is a magic, hallucinating ingredient. Rewiring a business around it takes colleagues who show one another what works, turning each new chalk your bobbins tip into shared advantage.

The cost is not just duplicated work. It is hesitation. Every person has to decide whether the meeting was the real meeting, whether the stated decision will survive the conversations afterward, whether everybody believes the plan or has merely agreed to stop discussing it.

We had none of that on March 3. The whole team understood the bet, believed the others understood it, and began.

Trust, spoken and inscribed

Trust is knowing I can tell you the expensive thing and neither of us will go around the other afterward. Affection, agreement, and long acquaintance are not enough.

Niall Ferguson wrote that money is not metal; it is trust inscribed. The $15 million check was trust accumulated outside the room and finally made visible. Before anyone wrote that check, we had to trust one another enough to make the wager. Milkana named the number and took responsibility for making it real. The check was trust inscribed. The wager was trust spoken.

The fundraise rested on decades of relationships, reputation, integrity, and people willing to show up for us.

The visible part took days. The invisible part took decades.

Trust compressed the diligence. It did not create the appetite. We were also standing where a large idea was about to land, and we had spent time in public saying things people could check. Decades of integrity got us the meetings. Conviction the market was about to share got us the term sheet.

When the clock is going 40x, there is no time to manufacture trust inside every transaction. Trust becomes the mechanism that compresses diligence.

At 40x, you need to know where a decision came from, who can see it, and whether it changed along the way. We are building those guarantees deep into SageOx.

I later sat down with Kumar Sreekanti to compare the raise from both sides. At Seattle's iconic Showbox, I told the public version of the same story: what looks like speed from the outside was decades of integrity underneath.

SageOx generated this mural automatically from my conversation with Kumar. The product drawing the argument for the product. Completely recursive.

Give the truth a handle

In the early days of Pixar, John Lasseter, Andrew Stanton, Pete Docter, Lee Unkrich, and Joe Ranft made a promise: no matter what happened, they would always tell one another the truth. They called the constructive criticism that followed good notes.

The name did not make them truthful. It gave a new colleague a way to recognize, invoke, and eventually inherit the pact already in the room. We give good notes here. Now being told that your reel isn't working can mean the system is functioning, not that the system has turned on you.

What a new colleague sees

At SageOx, every human-to-human, human-to-agent, and agent-to-agent conversation is shared into a common memory our coworkers read before they act. We record the engineering journey: every coding session, every wrong assumption, every rabbit hole, every breakthrough. We don't hide unfinished work because that's where the learning lives.

That is not naturally comfortable.

Emory Clark joined us as a designer in June. In her first week, she saw one of her own working sessions open on Ryan's screen. Nothing had gone wrong. The system was working exactly as designed. It was still a jolt.

Every new SageOx hire has to get used to the consequences of our radical transparency: your unfinished thinking is visible before you have decided what you think. So is ours.

The Ox CLI is open source. We keep the thinking behind it open too. Our discussions are recorded, our coding sessions land in the Ledger, and the false starts stay beside the shipped work. The work is shared closer to the speed of thought than the speed of a daily update. It is the next turn in working in the open.

This will change as we grow. A ten-person founding room is not a hundred-person company, and payroll, personnel matters, and customer confidences do not belong in a common view. The technology already supports group-based access controls and enforces those permissions at query time, so humans and AI coworkers see only the context they are allowed to see.

But access controls cannot choose the culture they protect. An early-stage startup in the AI era should spend at least as much time on its first team norms around trust as it does on the technology. You can add boundaries as the company grows. It is much harder to remove the habit of withholding the useful thing.

This is not for everyone

Radical transparency isn't a value on a poster here. It is how we work. Your unfinished work is visible. Your wrong assumptions remain part of the record. You are expected to say the same thing in the room that you say afterward, and you can expect the same from the founders.

This way of working favors people who are comfortable showing unfinished work, changing their minds on the record, and saying the same thing before and after the meeting. Disagreement is welcome; good notes stay specific and directed at the work. Context belongs to the team, and no part of the product is permanently somebody else's problem.

These peculiar tendencies make our work far more impactful, fast, and joyful. They are not for everyone. We are hiring our first ten, including a Product Engineer, Full Stack. If you'd rather show your work than protect it: jobs@sageox.ai.

Get the Ox CLI running, then drop us a note. We will happily walk you through our thinking.