Who Owns the Knowledge Inside Your Tools?

There’s a question I keep asking people, and the answer always unsettles them.

Who owns the rules inside the software your organization depends on?

Not who pays for the software. Not who administers it. Who owns the logic, the decision-making, the encoded knowledge that determines how the system behaves?

The answer, almost always, is: nobody in your organization. The vendor owns it. The platform owns it. The developer who left three years ago wrote it, and now it lives in a codebase that nobody fully understands. The knowledge that runs your operation exists inside a black box that you rent.

I’ve spent my career building these black boxes. Over two decades of enterprise software, national platforms, systems that millions of people use every day. And the thing that keeps me up at night isn’t whether the code works. It’s what happens to the knowledge once it’s inside.

Let me tell you a story.

I worked on a large-scale government platform in Saudi Arabia. The system handled critical national operations. The rules were complex, the edge cases were endless, and the domain expertise required to build it came from people who had spent their careers understanding how the real-world process worked. Regulators, inspectors, administrators, people whose knowledge was the foundation of the entire system.

We encoded that knowledge into software. We did it well. The system worked. But here’s the thing: once the knowledge was in the code, it belonged to the code. The domain experts who contributed their understanding couldn’t read the implementation. They couldn’t verify that the rules were still correct. They couldn’t modify the logic when regulations changed without going through a development team. Their knowledge had been translated into a language they couldn’t speak.

This happens in every industry. A hospital’s patient triage logic lives in an EMR system that nurses can’t inspect. A school’s enrollment rules live in a platform that administrators can’t modify. A food bank’s distribution priorities live in a database that the volunteers who understand the community can’t access. The people with the deepest knowledge of how things should work are the people with the least control over how things actually work.

Now multiply this by AI.

Large language models are trained on vast amounts of human knowledge. They can generate policies, write procedures, design workflows, make recommendations. They’re impressive. They’re also profoundly centralizing. When an AI model encodes knowledge, it encodes averages. It encodes the general case. It doesn’t encode the specific reality of your community, your organization, your context. It can’t, because it was trained on everyone’s data and nobody’s understanding.

The result is a world where local expertise gets flattened. Where the person who has spent twenty years understanding how water distribution works in their region is replaced by a system trained on how water distribution works everywhere. Where the nuances, the exceptions, the human judgment that makes systems actually serve people gets smoothed out into a statistical mean.

I find this genuinely frightening. Not because AI is bad, but because we’re building the most powerful knowledge systems in human history without asking who they serve.

The question of who owns knowledge isn’t academic. It determines who has power. When a company owns the rules that govern your operations, they have leverage over you. When a platform owns the logic that serves your community, they can change it without your consent. When an AI model absorbs your expertise into its weights, that expertise becomes theirs to monetize and yours to rent.

So what do we do about it?

I think the answer has several parts, and none of them are simple.

First, we need knowledge to be expressed in forms that humans can read, not just machines. Code is precise but opaque to non-developers. Natural language is readable but too ambiguous for machines. The gap between these two is where the real work is. Formal specification languages, structured rule systems, declarative logic, these are old ideas that deserve new attention. If the rules that govern your organization were expressed in a format that a policy expert could read and a machine could execute, the knowledge would stay visible even as the software changes.

Second, we need portability. Right now, switching software platforms means losing the institutional knowledge baked into the old system. The logic, the customizations, the edge cases you’ve spent years encoding, all of it trapped in a proprietary format. Knowledge should be exportable. If you leave a platform, your understanding of how your operation works should leave with you.

Third, we need to rethink the role of AI in knowledge work. The current model is extraction: AI absorbs human expertise into its training data and offers it back as a service. A better model would be amplification: AI helps humans express what they already know in structured, portable, verifiable formats. The difference is who retains ownership. In the extraction model, the platform does. In the amplification model, the community does.

Fourth, governance matters. Open-source software helps, but it’s not sufficient if the logic is still unreadable to non-technical stakeholders. Cooperative ownership models, community data trusts, public digital infrastructure, these are governance structures that can keep knowledge accountable to the people it affects.

None of this is easy. The economics push hard in the other direction. Centralized platforms are more profitable than community-owned tools. Black boxes are cheaper to build than transparent systems. Vendor lock-in is a feature, not a bug, if you’re the vendor.

But the cost of the current trajectory is real. Communities that can’t understand the systems they depend on. Organizations that can’t adapt their tools when reality changes. Experts whose knowledge has been absorbed into platforms that no longer need them. A world where the gap between those who control the code and those who live under it keeps growing.

There is an alternative. Systems where knowledge stays readable. Where communities own the facts, rules, and events that define how they operate. Where AI helps translate human understanding into shared system logic instead of replacing human understanding with statistical averages.

Where people remain part of the systems that shape their lives.

That’s not just a technology problem. It’s a human rights problem.

Osamah Alghanmi is the co-founder of Almadar, a software studio based in Ljubljana, Slovenia. He spent over two decades building enterprise systems, including national-scale platforms in Saudi Arabia. He writes about knowledge, power, and the question of who controls what machines know.