LMIM OS is a local AI assistant in one file. LLM chat, document Q&A, voice, WhatsApp, web scraping, scheduling, and dev tools — all offline. No cloud, no API key, free forever. Unlike thin chat wrappers, LMIM bundles the full stack: inference, voice, RAG, messaging daemons, and sandboxed workspace. Built for an English school in Tijuana where student data couldn't leave the building. RAG • Workspace • Voice • WhatsApp • Scraper • Contacts • Calendar • CUDA

LMIM OS v2.1 'Tezcat · Sharpened' — Local AI with RAG, workspace sandbox, web scraper, contacts, voice, CUDA acceleration, bundled model. Linux + Windows Live Now. No setup. No subscription. No API key.

Hey Product Hunt 👋 I'm Andrés, building LMIM OS . This started as a tool for my English school — I needed AI that could handle student data without sending it to the cloud. The existing options were either too technical or too privacy-compromised, so I built my own. The product has been live for ~3 months with 1,000+ organic downloads — no ads, no launch, just people finding it and installing it. Today is the first time I'm actively putting it in front of people. What's crazy is the build complexity. Getting llama.cpp, whisper.cpp, Piper TTS, Baileys (WhatsApp), and sentence_transformers to all run in a single frozen Python binary wrapped in an Electron shell — across Linux AND Windows — took about 50 build iterations. The torch/distributed import problem alone was a multi-day debugging session. I'd love feedback on: What features would make this a daily driver for you? What's the biggest friction point in current local AI tools? Would you pay for a "team mode" version with multi-user support?

LMIM OS Lean Mean Inference Machine

Local AI with RAG, voice & WhatsApp. Open it and it works.

Local AI with RAG, voice & WhatsApp. Open it and it works.

LMIM OS v2.1 'Tezcat · Sharpened

Previous LMIM OS Lean Mean Inference Machine Launches

Previous LMIM OS Lean Mean Inference Machine Launches