A local AI assistant can be set up on a Windows computer using the Ollama tool, which allows running language models without sending data to external cloud services. The setup process requires no registration, API keys, Docker, or Linux knowledge.
To start, the user installs Ollama from the official website and runs the command 'ollama run deepseek-r1:8b' in the Windows command prompt. The model file size for this version is 5.2 GB, and the download speed depends on the user's internet connection.
The author tested the model on a laptop with 16 GB of RAM and an 8 GB graphics card. For systems with lower specifications, a lighter version of the model, deepseek-r1:1.5b, is available. The model provides responses to coding and SQL queries locally.
While local models offer privacy for sensitive data, they are not direct replacements for cloud-based AI services. Performance depends on hardware capabilities, and smaller models may produce inaccurate or hallucinated information that requires manual verification.
