Hacker News: dimitrismrtzs

New comment by dimitrismrtzs in "Incident Report: May 19, 2026 – GCP Account Suspension"

dimitrismrtzs — Sat, 23 May 2026 02:38:23 +0000

stories like this are why i self-host most things on proxmox instead of depending on a single cloud provider. Ok I have to do the maintenance but at least this way no one can suspend my entire stack with not even an explanation.

New comment by dimitrismrtzs in "How Claude Code works in large codebases"

dimitrismrtzs — Mon, 18 May 2026 18:02:17 +0000

So if 100k codebase is considered large, this is how i do it. I keep a documentation folder, one for the backend one for the frontend and i try to keep these as up to date as possible. So every time i start a new feature or try to fix a bug, i tell claude to read what it needs from the documentation first and mention the thing we need to focus on. I dont know if the focus on documentation is even needed that much but with this flow i have seen claude sucessfully do some pretty complicated stuff and not a problem with exploring the codebase.

New comment by dimitrismrtzs in "Granite 4.1: IBM's 8B Model Matching 32B MoE"

dimitrismrtzs — Thu, 30 Apr 2026 13:32:40 +0000

The 8B class closing the gap with 32B is the real story of 2026 for anyone running models locally. I've been using smaller models for agent tool-use and the progress this year is real.

The gap that still matters most isn't intelligence — it's consistency on structured output. When you chain 5+ tool calls in sequence, even a small per-call reliability difference compounds fast. Would love to see Granite 4.1 benchmarked specifically on multi-step function calling rather than just general benchmarks.

New comment by dimitrismrtzs in "Ask HN: What would you do with an AI model capable of continuous learning?"

dimitrismrtzs — Fri, 10 Apr 2026 01:45:40 +0000

Honestly with the latest mythos news I am a bit concerned about when this happens. But I would test it as a CEO.

New comment by dimitrismrtzs in "Ask HN: Is building a company around an open-source AI agent platform realistic?"

dimitrismrtzs — Tue, 07 Apr 2026 23:11:14 +0000

Do you have any idea about the right choice for the license?

Ask HN: Is building a company around an open-source AI agent platform realistic?

dimitrismrtzs — Tue, 07 Apr 2026 06:40:12 +0000

I've been running a Proxmox lab with around 40 services for a while now. Over the past year I ended up building an AI agent platform that manages my infrastructure and handles business operations — documents, phone calls, scheduling, monitoring, that kind of thing. It wraps Claude Code CLI as the execution engine with custom MCP servers on top.

The whole thing started as internal tooling for a startup I was working on but honestly the platform ended up being way more useful than the startup itself. I'm seriously considering open sourcing it and going full time on it. The way I think about it is something like Home Assistant but for business operations — you self-host it and AI agents handle your operations.

My concern is sustainability though. I can't realistically maintain something like this on the side — if I do it I need to go all in. Has anyone here built or seen a real business around an open source tool in the AI space? What license would you go with? I've been looking at Modified Apache 2.0 (what Dify uses) vs BSL but honestly not sure what makes sense.

Would appreciate any advice from people who've dealt with this.

Comments URL: https://news.ycombinator.com/item?id=47671502

Points: 1

# Comments: 2