The Cyborg Manifesto

The biggest social question of our time is what happens to people as AI gets better at work. Our answer: AI should not replace humans, but empower them.

Most AI tools today make a small group of engineers faster and leave everyone else watching from the sidelines. Cyborg is built to do the opposite. We bring every person on a team, technical or not, into the loop of building together with AI agents, so the designer, the support rep, and the new hire with a bright idea can all contribute their unique insights into real product. We call it Agents working, People guiding.

We know from experience that great ideas can come from anyone. When building no longer requires writing code, who gets to be a builder stops depending on title, training, or background. That is the future we are working toward: one where AI widens the circle of people who create, rather than shrinking the number of people who work.

That is the manifesto. The rest of this post is what we mean by it, and how it shapes what we build.

The question

Every company is asking some version of the same thing right now: AI can do more of the work each month, so what do we do with our people?

There are two answers. One treats AI as a way to need fewer people: keep the output, cut the headcount. The other treats AI as a way to get more from the people you already have: keep the team, and let each person do things they could not do before.

Both are possible with the same technology. Which one happens is a choice, and a lot of that choice gets made in the design of the tools. A tool that only one kind of person can drive concentrates the gains in that kind of person. A tool the whole team can drive spreads them.

Who gets left out today

AI made each person on a team faster. It did not make the team faster.

The coding agents of the last two years are remarkable, and almost all of them are built for one engineer at one terminal. The engineer gets faster. The product manager, the designer, the ops lead and the support rep still write a ticket and wait, the same as before. Their knowledge of the customer reaches the product secondhand, if it reaches it at all.

That is a strange place to end up. The people closest to the customer are the furthest from the tools that build the product.

Agents working, People guiding

Cyborg is a shared workspace where a team and its AI agents build together. Anyone on the team can start a mission in plain words. Anyone can jump in. The agents do the building and check each other’s work, and a person is pulled in when the work needs a judgment call.

The division of labor is in the name. Agents are good at the work: writing, testing, checking, doing it again. People are good at the guiding: knowing what the customer meant, what good looks like, and which of two right answers fits this company. As agents get more reliable, checking their work matters less and that judgment matters more.

So the scarce thing on a team stops being the ability to write code. It becomes taste, judgment and knowledge of the customer, and those were never limited to engineers.

Great ideas can come from anyone

The support rep knows which bug costs the most customers. The designer knows why the flow feels wrong. The new hire sees what everyone else stopped noticing months ago.

In most companies those insights have to survive a ticket, a prioritization meeting and a queue before they become product. Many do not. In Cyborg, the person with the idea can start the mission themselves, see it built, and guide it until it is right. The best idea in the room wins, no matter who had it.

What this commits us to

A manifesto is only worth something if it costs you decisions. These are ours.

• People are free. We charge an organization for compute, never for how many people join. A price per seat is a reason to leave someone out, so we do not have one.
• Everyone sees the work their own way. A designer, a founder and an engineer do not need the same screen. Each role gets a view built for how it works, on the same missions.
• No one asks for permission. Anyone on the team can launch and direct a mission, not only engineers.
• People stay in the loop. Agents pull a person in when a decision is theirs to make. We are not building a way to take the team out of the work.

The future we are working toward

We do not think the measure of this technology is how many people it lets a company do without. We think it is how many people it turns into builders.

When building no longer requires writing code, the number of people who can create something real goes from a few on each team to everyone on it. That is the future we are working toward: one where AI widens the circle of people who create, rather than shrinking the number of people who work.

Agents working, People guiding.