The author describes running a local coding agent using Oh My Pi with Qwen3.8-27B on two RTX 3090s. Key optimizations include adjusting thinking budgets, token limits, and subagent concurrency to achieve practical inference speeds. Local setups offer privacy and cost predictability but require careful tuning and accept slower inference compared to hosted frontier models like Claude or GPT.
Otis is an open-source AI agent that runs local and hosted open-weight models with privacy-focused design. It automatically recommends and downloads appropriate models based on hardware and supports llama.cpp, Ollama, LM Studio, and Nvidia PAIR.
Mees AI is a privacy-focused offline AI workspace featuring dual modes: Arena for multi-agent debates and brainstorming, and Journal for private reflection and mood tracking. All processing runs locally on-device with no cloud servers, subscriptions, or APIs required—available as a one-time lifetime purchase.
BiNeuron is a local AI platform that analyzes code requests and generates production-ready code across 25+ programming languages. It combines natural language processing, OCR, and adaptive model selection with support for multiple file formats and automatic hardware optimization.
The author built Deckard, a Chrome extension that uses local AI models to automatically detect AI-generated text in the browser without sending data to third parties. While existing open-source models perform worse than commercial services like Pangram, they are accurate enough for practical use and will improve over time.