From one prompt to a deployed app: a self-hosted, sandboxed AI software factory
A self hosted AI software factory
A developer builds an almost fully self-hosted, sandboxed, agentic software factory on his home server. Using Coolify, Forgejo, Hermes, and Firecrawl, he creates an isolated environment where an LLM can autonomously handle the entire SDLC—from research and coding to CI and deployment—without giving it root access to his main machine. The system even provisions SSL certificates via DNS-01 challenges, keeping services private on his tailnet. The result: a single prompt led to a fully deployed calorie-tracking app with tests and CI, all without further human intervention.
It’s not a complicated app, obviously, but it went from a paragraph to tested, deployed software and handled all the boring bits in between.
- ashu1461
In such systems, producing code is the easy part, verification is hard. Verification via test cases just feels like the same agent validating its own assumptions.
Wondering what the actual verification loop looks like once you start taking these systems to production.
- RajT88
So - apparently it's not fully self-hosted, since I don't see a GPU.
I'm interested in hearing from folks who are hosting their own GPU to run coding models. So far my own results are... not great. Seems like frontier models are needed via the big providers?
- copemaxxxing
People who build software factory are in for a disappointing future. Spoken as someone who uses AI day to day for job and personal projects, I've came across a few super gnarly bugs already, ranging from frontend React apps (yes, believe it or not, frontend is far from solved) to embedded that Claude can't solve. Those bugs are feature breaker. My primary skillset is fullstack/frontend leaning, not embedded. Granted, I might be out of depth in embedded/RTOS but I am very qualified to do frontend. Yet I found bugs in those areas that Claude Opus high can't even solve for hours.
I pointed Claude at those for hours and hours, didn't fucking work, ended up creating more and more slop. Until I stepped in and fixed it in a few lines of code. Thankfully with all my AI assisted work, I took time to understand everything so I was able to debug and fix.
The problem for software factory is, that up to a certain point, you'll encounter some nasty bugs that are really important for your features, but you can't ship it because while the AI deals with the 90%, turns out that last remaining 10% are the one that make or break your project. And you need to deal with that remaining 10%. But how can you deal with the remaining 10% if you don't understand that 90% or majority of it?
Also, don't listen to influencers like Kent Dodds or Uncle Bob or Steve Yegge, who advocated for software factory walk away from the code kinda thing. They are already rich, and they sell teaching content, but most im […]
- codazoda
I was recently inspired by another article here to start my own. My skills are written and tested, the factory has built the first test project, and I’m setting up the final machine to run it. Here’s my initial post about my motivation and early plans plus some follow-ups, and there are more to come.
- Kinrany
Is there a name for the infrastructure stack that is designed to be operated by a human instead? Spawn agents each in a separate VM with the repo checked out and the tooling set up, allow them to spawn subagents in the same manner, but pull their changes from their branches (set up as remotes in your own dev environment) and merge into trunk yourself?