The Encumbrance

By Ricardo Vega

---

My friend Mr. J avoids AI wherever he can — outside a few limited, deliberate exceptions — and his reason is better than most of the reasons people give.

He isn't worried about robots. He's worried about ownership. His argument is that AI is a capital multiplier: it makes labour cheaper for whoever already owns the means of production, so the gains flow upward and the people whose work got cheaper get nothing. Worse than nothing: some get displaced, and the ones “lucky” enough to keep their jobs get saddled with more of the work. Every productivity gain in his lifetime has been captured by someone who wasn't doing the work. He sees no reason this one will be different, and he's largely right about the default.

I've argued with him about it for a year, and here's what I eventually conceded: his critique isn't wrong, it's incomplete. It's a description of what happens when you aim the technology at a particular target. Change the target and the mechanics change.

He also changed my mission statement. This organisation used to be about technology and homelessness. It's now about using technology — not AI specifically, all of it — to make people's lives better. AI happens to be the catalyst available right now. That widening came out of arguing with Mr. J, and it's the more honest description of what I was doing anyway.

Then I got a contract with the Texas Education Agency, and the abstract argument turned concrete.

I met people who were faintly ashamed of where they worked — who'd say the name of the agency the way you'd confess to something. Government work, bureaucracy, the machine. Meanwhile they were writing code that teaches Texas children.

One of them I'll call Mr. R, a developer I worked alongside, and he was brutally honest with me in a way I've thought about ever since. Not that he hated it. It was more resigned than that: here I am, back at this job, because what else am I going to do. A job as the thing that happens to you when the alternatives run out.

And I said to him more or less what I'm about to say to you. Mate — you're changing kids' lives. That's a badass thing to be doing with a Tuesday. Why does it not feel like one?

Because the connection between the work and the outcome had gone missing somewhere. He wasn't wrong about his day. His day was tickets and requirements and an API. The children were three abstraction layers away and nobody had ever bothered to draw the line back.

The second thing happened about a year before that, and I didn't understand it until I was inside the agency.

I was at one of my sons' games. The opposing school had a teacher there — he'd taught two of my boys maths — and I got talking with his wife, who I'll call Mrs. A. She teaches special education. When I asked what she was doing now, she said she was back in special ed, and something in how she said it didn't match the words. Not unhappy exactly. Weary.

Which confused me, because from the outside that's the job. You change children's lives. I can't think of a better use of a working day.

So I asked her what makes it hard. Not what's wrong — I assumed the kids were the reward, and I wanted to know what was sitting on top of the reward.

She said the regulations. Keeping up with what the agency requires.

Not the students. Not the behaviour, not the parents, not the pay — and special education teachers are not paid what they're worth by any measure. The hard part of her job was documentation. The compliance layer between her and the thing she went into teaching to do.

Eight months later I was contracting for the agency that produces that layer.

And that's where Mr. J's objection and Mrs. A's answer meet — the abstract and the concrete — because they're the same argument seen from two ends.

Aim AI at the part of a job someone loves, and Mr. J is right. You've built a device that replaces labour that was being paid for, the savings go to whoever owns the device, and a person who found meaning in that work now does something worse for less. That's most of what's being funded right now, and there is no reason to celebrate it.

Point it at Mrs. A's pain point instead — the encumbrance: the paperwork, the compliance narrative, the form that exists so one agency can prove something to another — and the arithmetic inverts. Nobody's job was that. Nobody chose it, nobody's identity is in it, nobody is paid extra for being good at it. It's friction, and every hour you remove from it goes back to the human doing the part only a human can do.

Mrs. A doesn't get replaced. Freed from the filing and the bureaucratic busywork, she gets to focus on what she went into teaching to do: helping her students become the best versions of themselves they can be. In short — Mrs. A gets her Tuesday afternoon back.

That's not small, and not just to her. The value created doesn't bubble up to some nameless process outside her purview. There's no owner capturing a margin on her recovered afternoon. The gain lands on a child in a special education classroom in Texas, and there's no mechanism by which it flows upward. That's what Mr. J's critique is actually pointing at: not the technology, but whether the value it releases has somewhere to be captured.

Which is the same test I applied to the food. Surplus in a dumpster is value nobody is capturing — it's being destroyed. Route it and you create value out of waste rather than out of somebody's wage. The compliance burden on a special education teacher is the same shape. Pure friction, owned by no one, benefiting no one. Removing it takes from nobody.

So here's the test I've settled on, and I owe it to a man who barely touches the tool. Whose work does it steal, and who benefits from the heist? If it takes work someone wanted to do and the value pools with whoever owns the system, be suspicious. If it takes work nobody wanted — if it breaks down a barrier between someone and the thing they love — build it.

I am one person with a contract and an opinion. Mrs. A's problem is not one teacher's problem — it's every special education teacher in every district in this state, spending evenings on documentation instead of on the work or on their own families. And the thing standing between them and relief is not a hard technical problem. It's an obtuse API, a pile of requirements written in agency dialect, and nobody with the time to sit down and translate one into the other.

That is an extremely solvable problem, and it is exactly the kind of problem people like me and Mr. R solve for money every day for companies that matter considerably less.

So this is a call to the nerds — and this is the part that's actually a request.

If you write code, and you understand — or want to understand — how the education system actually works underneath, there is a genuinely worthwhile thing to build here. Not a startup. Not a platform with a pricing page. Tooling that takes the compliance layer off a teacher's evening and hands it back to her.

What I think it looks like: someone in each district who knows that district's requirements and quirks, someone at each school who knows what's actually being filled in and when, and a group of developers working with them rather than at them. Multi-tenant, so what's built for one district serves the next. Shared in the open, so when somebody works out how to satisfy a requirement correctly through that awful API, everyone gets it instead of forty districts solving it separately and badly.

That last part is the whole point. Right now every district is paying, separately, to lose the same fight. The knowledge exists — it's just scattered across people who've never been in the same room.

And to the developers who feel like Mr. R: you are writing code that teaches Texas children — equipping the next generation not just to succeed but to shape the world they'll inherit. That is a real thing, and it got buried under tickets. This is a way to feel it again — to see the line from the thing you wrote to the afternoon a teacher spent with a kid instead of a form.

Nobody's going to fund this properly. It doesn't need much funding. It needs a few people who'd rather their Saturday counted for something.

If that's you, come find me.