The World Is Changing Fast

En Parlant~ is a fork of En Croissant, the open-source chess study tool by Francisco Salgueiro. I added text-to-speech narration — five providers, eight languages, translated chess vocabulary — so you can hear annotations read aloud while you keep your eyes on the board. For me, it helps me study chess. I can load an annotated game, and step through it while the voice is narrating. This helps me focus on what is happening on the board versus flipping back and forth to the .pgn file. I find it helpful, others may not. I wanted to put this into the game, but it wasn’t the direction En Croissant wanted to go,and that’s how open source works. Without En Croissant, this project doesn’t exist.

This is a note about how it got built, and why I’m being upfront about it.

Kasparov had the right idea

I am old enough to remember May 1997. Deep Blue beats Kasparov. The world was agape, and everyone thought the human brain just became obsolete. Kasparov stormed off, but he didn’t quit chess — that part gets left out. He kept playing, studied what the machine did, and then came up with Advanced Chess: humans and computers playing together. A strong human paired with a machine could beat the machine alone. He called it the centaur model.

Nearly thirty years later I am adding to a chess app with an AI. We have to really take a pause and understand what that means.

How this was built

This fork was built with Claude Code, Anthropic’s AI coding assistant. Rust commands, React components, the whole TTS pipeline. Pair-programmed, human and AI. I’m not going to pretend otherwise.

There’s a difference between AI slop and co-development. AI slop is typing “make me a chess app” and shipping whatever comes back. This is co-development. I’d say “no, that’s wrong, here’s why.” It’d come back with “fair, but what about this edge case.” I brought the direction. It brought the speed.

Building from a solid start

Here is the rub I want you to think about. Francisco Salgueiro built En Croissant over years. Late nights debugging Tauri quirks. Careful UI work with Mantine. A PGN parser that actually works. Engine integration, database management, puzzle support. He built something he cared about and it shows. He and his team spent years making hard decisions, learning the framework, shaping the architecture through hundreds of commits. Everything an AI adds sits on top of that. An AI helped me move fast, but moving fast on top of someone else’s years of effort isn’t the same as doing what they did. I don’t have a tidy way to square that. I just think it’s worth saying.

What it means, practically

The chess community gets a better tool because AI coding assistants exist now. The gap between “I wish this app could narrate moves” and “it narrates moves in eight languages” went from many months to a weekend. That’s real.

It’s also happening everywhere. Software that took teams of developers takes one now. Some of that is great — a solo dev with a good idea can actually ship. Some of it is unsettling. When the cost of building drops this far, what separates good software from bad? I think it comes down to taste and judgment and actually caring about the result. Those are still human things. But I won’t pretend the shift is painless for everyone.

Why chess

Chess is a mental refuge for me. I came back to it after years because I can park all the devices, step away from the machine and sit in a park in front of a wood board with 64 squares, 32 pieces, no chess engine, no app, and come up with whatever my brain can do with them. I like to train with all the tools, but when it comes time to playing, it is just me. As it should be. And that’s a good thing. AI is going to do a LOT of thinking for us in the future, and I just want to have a “mental gym” where I can go to and get away from all that.

About honesty

I’ve seen repos with thousands of lines of AI-generated code and zero acknowledgment. I get why — there’s a stigma. People hear “AI-assisted” and think lesser.

But in my day job, I’m an airline captain. Thirty-nine years flying and counting. My cockpit is full of automation — autopilot, auto-throttle, fly-by-wire, GPS, synthetic vision. Nobody asks me if I’m “really” flying the airplane. The tools don’t diminish the skill. They change what the skill is. Judgment, situational awareness, making the right call when nothing is clear. That part is still mine.

Software is going through the same shift aviation went through decades ago. The hand-flying purists will always exist, and I’m one of them — if you’re on one of my flights, you might catch me hand-flying the whole thing, takeoff to touchdown. Because I can, and because that skill matters. But I also know when to let the automation work.

This project was built AI-assisted. It really is an inflection point that you should sit and think about.

Open process

This repo includes the actual Claude Code workflow document — what the AI knows, what it’s told, how sessions work, where I draw the line. If you’re curious what AI-assisted development looks like in practice, it’s all there.

Wrapping up

If you build with AI, be honest about it. If you’re building on someone else’s work, say so.

Deep Blue beat the world champion in 1997 and chess didn’t die — it got bigger. More players, more study, more depth. We’re at a similar point, and this is just the beginning. The tools have changed. The questions about craft and honesty haven’t.

Be safe out there.

-Darrell

---

En Parlant~ is a fork of En Croissant by Francisco Salgueiro, built with Claude Code by Anthropic.