The myth of values as a competitive advantage

Why values decay

Companies rarely drop a value because the people in them turn bad. They drop it because keeping a value costs something every day, and skipping it costs nothing this quarter. Every exception is reasonable on the day it is made. Add them up over a few years and you get a company nobody chose. It is the same pull we described in Why Castor: staying different takes work every day, the way a dam needs patching every day.

So the values that last are the ones arranged to pay for themselves. Kelly Johnson knew this about pay. At Lockheed's Skunk Works he gave raises to whoever supervised the fewest people, because in the company's main plant raises went to whoever supervised the most, which gives every manager a reason to grow a team.³

The test

By the end of that August morning we had a first test. A value is an advantage when it sets you apart inside a market you already compete in. If all it does is shrink your market, it is a cost, and costs worth paying should be called costs and paid on purpose. Every value loses you somebody. One of us thinks that is fine: a million people who care about what you refuse to do can be enough, if you go deep enough for them. So the question is whether a value wins you more of the people you want than it loses.

In 1945 Enrico Mattei was put in charge of Agip, Italy's state oil company, with instructions to close it. He built it into ENI instead. The oil business then belonged to the seven companies he named the Seven Sisters, and they had settled on giving a producing country half the profits. In 1957 Mattei offered Iran three quarters. To Iran and Egypt he also offered to carry all the risk of exploring: if there was no oil, they paid nothing.⁴⁵ A latecomer with little oil of its own got into a closed market by giving the other side the better deal.

Mattei is also alleged to have bribed politicians and journalists on a large scale, and he said he used political parties like taxis: get in, pay for the ride, get out.⁴ We keep him in the story for that reason. The better deal did its work whatever was in the heart of the man who offered it.

You could read both our examples as something plainer. Hinge's slogan is good marketing, and Mattei's deal was hard bargaining by a man who needed a way in. We think that is right, and it is the uncomfortable part of the idea. From the outside, a value that pays looks exactly like strategy. The difference only shows on the day it stops paying.

We think more values will pass the test from now on, and this is a bet, not a finding. With so many people shipping small things fast to raise money, and machines that can produce a passable version of almost anything, being recognisably different looks worth more than it used to.

Whose values

The other objection from that morning was sharper. Values like these put our morality above the user's, and there is no morality everyone shares. People are entitled to want things we would never build, and deciding for them is its own kind of arrogance.

Our answer, so far, is to keep our values about what our tools will and will not do on someone's behalf, and out of what people ought to want. Our email agent reads and drafts everything and cannot send. Some people would rather have an agent that sends, and they will pick another tool. For everyone who has watched an agent do something they never asked for, the limit is the reason to try it.

Ours, dated

Writing a value down in public is cheap, and breaking it afterwards is not, which is the whole point of doing it. These are the ones we hold as of 7 October 2026, with where you can check us:

• Anything an agent does in your name that is hard to undo waits for your own click. Gaddi, our browser server for agents, holds actions like publishing until you approve them with Touch ID; our email agent drafts and cannot send.
• You decide which agents can see which parts of your data.
• Leaving is easy. Most of what we publish runs on your own machine, so your data stays with you.
• What we build for ourselves and publish, we publish under an open licence that anyone can fork.

They are also the direction we steer by while we explore, and the thing to hold us to while we take our time choosing what to build.

None of them has cost us anything yet. The first time keeping one loses us a customer we would have liked to keep, we will write here which way we went.