# Why we open-source — Andrometiq

We open-source YRP and the software we build. Both honest reasons — the one that is good for everyone and the one that is good for business — and a straight answer to the strongest case for closing code: the data moat.

Canonical: https://andrometiq.com/open-source/

[Home](/) › A position

A position

# Open, on purpose.

“You spent years building this. Why would you give it away?” It deserves a straight answer, not a slogan. There are two real reasons. One is good for everyone. The other is good for us — and we would rather tell you that one plainly.

The reasons

## Two reasons. Both real.

Good for everyone. There is more software in the world than ever, and more of it than ever is slop — generated fast, shipped unlived-in, quietly abandoned. Ours earned its shape on a working garment floor, problem by problem. Closed, that understanding helps one company at a time, for a fee, forever. Open, a problem gets figured out and fixed once — and stays fixed, for everyone who comes after.

Good for business. Open source is also, plainly, commercially rational — and saying so out loud is the point. The business motivation is reputation and expertise. Everyone you meet claims expertise; almost nobody can let you check. Opening the work makes the claim checkable — by you, by your developers, by anyone, without our permission.

> We would rather be verified than believed.

The argument

## Closing the source now buys nothing.

Keeping source closed was always a trade. You accepted real costs — buyers who cannot inspect what they are buying, lock-in fears priced into every deal, every flaw fixed once per vendor behind every wall — and in exchange you got one benefit: a competitor could not copy your product.

For business software, that benefit has collapsed. A business system carries no hidden magic: its behaviour is its specification. Anyone who works in one for a week knows what it does. And with AI coding, anyone who knows what a system does can have a working equivalent built — quickly, cheaply, without ever seeing a line of your source. The wall still stands. There is simply nothing left behind it that the wall protects.

Because the hard part was never the code. The hard part was knowing what to build — which screen a floor will actually accept, which control actually stops the leak, what to refuse to build at all. That knowledge does not live in the source; it lives in the people and the practice that produced it.

So the trade now runs one way: the costs of closing remain — carried mostly by your customers — and the protection is gone. Closing the source is a cost with no return.

Steel-manning the other side

## We went looking for the exception.

An argument this convenient for us deserves an honest search for where it fails. These are the strongest cases we found, each answered on its merits.

### The data moat

In the age of AI, whoever holds the data wins. We think that is largely true — data compounds, tunes products, trains models, and cannot be recreated by a competitor at any speed. But look at what it is made of: data, not code. Notice that the companies sitting on the deepest data open their code the most freely — frameworks, tooling, even models — while the data stays locked. Opening the software never opens the data. And in our business the data was never ours to moat: what our systems record on your floor is yours. So the strongest exception is real — and it is an argument about data. It gives nobody a reason to hide code.

### Security by secrecy

Hiding code hides flaws from the people who would report them, not from the people who would probe for them — attackers work from behaviour, not from source. A hidden flaw is not fixed; it is waiting. We prefer the version where being seen keeps us honest.

### The licence business

If the licence key is the revenue, openness looks like ruin — and for that model, it is. But that model was already ending: a product that can be rebuilt from its behaviour cannot charge rent on its behaviour for long. The durable revenue was always the work around the software.

### The true trade secret

A company whose entire edge is one extraordinary algorithm has something a wall can protect, for a while. We concede that case narrowly. It is not this case: business software has no such secret. Everything it does, it does on a screen, in front of its users, all day.

### What stays closed

Your particulars. Your data, your configurations, your documents, your numbers — everything that makes a deployment yours is confidential by default and stays that way. We open the product. We do not open you.

> One real moat survives the search — and it is not made of code.

The question under the question

## “If AI can build anything, what are we paying you for?”

### Curation

A thousand screens are possible; a floor will accept one. Knowing which one — before a year of rework proves it — is the skill, and it was learnt the expensive way.

### Domain judgment

Knowing what to build, what to refuse to build, and where the leak actually is. The prompt is not the hard part; knowing what to ask for is.

### Accountability

Code is cheap now; a name on the outcome is not. Someone has to stay until it runs, and answer for it after.

### Operations

Software is an event; running it is a practice. Real volume, real consequences, every day, until the fix holds.

AI made writing code cheap. It made knowing what to write more valuable, not less — and it made “we will stand behind it” the scarcest sentence in the industry.

And one more thing

## The gesture that fits the moment.

There is an old line in computing: information wants to be free. Intelligence, now, has joined it — the judgment that used to be scarce is a prompt away, for everyone. We are not interested in fighting that world. Opening our work is how we choose to live in it.