About

Tech-kutak is an independent blog about AI, local large language models, and practical technology. I write about what I'm learning building with LLMs - the tools, the techniques, and the real-world applications.

Why This Blog Exists

I started Tech-kutak in March 2026 because everything about AI was moving fast. New models, new tools, new capabilities every week. I needed a place to document what works - and share it with others on the same journey.

What you're reading isn't theoretical. Everything here comes from actual hands-on experience running models locally, building AI-powered workflows, and testing approaches in my home lab.

What I Cover

  • Local LLM Setup - Getting models running on your own hardware
  • AI Coding Assistants - Prompt engineering, agents, automation
  • Model Comparisons - Real benchmarks, not marketing claims
  • Practical Tutorials - Step-by-step guides that work
  • Industry Trends - What's actually shipping, not vaporware

Who This Is For

  • Developers exploring local AI
  • Engineers building AI-powered applications
  • Tech enthusiasts curious about LLMs
  • Anyone wanting to understand what's possible today

If you're looking for fluff or clickbait, this isn't the site. I write what I'd want to read.

The Stack

This site runs on:

  • Ghost (self-hosted)
  • Proxmox for infrastructure
  • Ollama/lm Studio for local inference
  • Various open-source AI tools

Get In Touch

Questions, feedback, collaboration:

Email: ivan_horvatic@hotmail.com

I read and reply to all emails. Usually within 24-48 hours.

Disclosure

This site is independently owned and operated. I don't take sponsored posts or paid reviews. Any affiliate links are clearly marked.

When I review products or services, the opinions are my own - whether I paid for it or got it free.

Privacy & Data

I don't track you. No analytics cookies, no third-party trackers.

The only data I collect is your email if you subscribe to the newsletter - and that's only for sending you new posts.

See our Privacy Policy for details.


Building tomorrow, one prompt at a time.