Hello, Keystone

· meta · vision

Why I'm building an AI-native indie game studio in public, and what this devlog will cover.

I’m building an AI-native indie game studio. The codename is Keystone. The plan is to ship a desktop app that takes a serious indie designer from “interesting idea” to “commercial Steam title” with an AI studio crew — specialized agents for design, engineering, art, audio, narrative, QA — executing under human direction against a single design intent graph.

This devlog is where the work happens in public.

Why public

Indie game development is famously hard alone. AI-native tools change the math, but only if the tools are honest about what they can and can’t do — and most of the public discourse about AI in gamedev right now is either breathless hype or reflexive skepticism. Neither is useful to an indie trying to decide what to use, what to build, or whether the whole premise is real.

The most useful thing I can offer is the build process itself: what shipped this week, what broke, the dead ends, the recalibrations, the moments where the AI made the work obviously faster and the moments where it didn’t. The progress matrix on the home page is the same artifact I use internally to track Keystone’s platform components from research to 1.0 — it’s not a marketing surface.

What you’ll see here

  • Weekly-ish posts from the build floor. Some weeks more than one, some weeks none. No content treadmill.
  • Real session transcripts from the Claude Code sessions where the work actually happens. Polished for readability, not for theater.
  • The progress matrix, updated the day work lands. If a column moves, there’s a post about how.
  • Honest tradeoff writeups — when an approach doesn’t pan out, you’ll see the autopsy, not a quiet revert.
  • No funnel. RSS and email get the same posts on the same day. No paywalled tier, no “premium subscribers” content, no engagement-bait.

Who this is for

Game designers, developers, indies, hobbyists — anyone interested in what an AI-native studio actually looks like in 2026 when you’re not trying to replace humans, but to give a single human director leverage they couldn’t otherwise have. If that’s you, the RSS feed and the email list are at the bottom of every page.

More soon. The next post is about the design intent graph — the wedge that everything else in Keystone hangs off — and why the whole thing falls apart without it.

Follow the devlog

New posts in your inbox. Same posts, no funnel.