Demo: Building a project using local AI (llama.cpp + DeepSeek/Harness)

In this video I walk through my workflow for developing quick projects using local AI models, without relying on cloud services. As an example, I build a tic-tac-toe game in HTML5 in real time, comparing two tools:

llama.cpp (web interface) for local code generation

DeepSeek via Harness, with close-up shots of the development process

The goal is to show how you can prototype and build functional software using AI that runs 100% on your local machine, no API costs and no dependency on an internet connection — ideal for MVPs, micro SaaS products, and fast proof-of-concept work.

If you're looking for a specialist in local AI environment setup (Ollama, LM Studio, llama.cpp), n8n automation, or AI-assisted development for small to medium projects, I can help you define the right stack and get it running.

Full video: https://youtu.be/eYOiZU5LDaU