BonzAI – self-sovereign, local LLM inference in the browser
Local AI suite tangled with crypto wallets and token rewards.

Local AI suite with crypto tokenization, but Pinokio does the inference part better.
Local AI enthusiasts, crypto-curious users
Pinokio · LM Studio · Ollama
I started building BonzAI two years ago, as an attempt to fully replace my online AI stack with an offline, sovereign alternative.
Back in the days, I had the intuition that open weights model would stand the comparison with private AI labs, and that the commoditization of inference was the future.
When I was finally able to switch from ChatGPT/Claude to local LLMs, midjourney to Flux/SDXL/Z-image, Eleven Labs/Suno to Qwen-TTS/Ace-Step, and Sora to LTX, I implemented:
- The ability to create datasets and fine-tune LoRAs;
- An experimental 3D objects / Game generation engine;
- Support for OpenClaw / Hermes Agent;
- A way to tokenize AI artifacts into tradable collectibles and share BonzAI profits with our ecosystem participants;
- Arguably the most immersive AI roleplay experience on the market that combines locally-installed models into multi-modal chat experiences with "BonzAI Companions";
- A marketplace that applies the BonzAI Companions approach to domain-specific application (e.g. Mediation Forge, that generates on-demand meditation session using LLM, TTS, image generation and Ken Burns effects.
For now, BonzAI is a desktop application available for MacOS, with releases planned for Linux / Windows later this month.
I think Intelligence should be fully owned, not rented (à la Sam Altman).
Hope you like it.
It's a self-funded effort that might help move the needle a bit (these OpenClaw token bills are pretty awful, currently).
W
Local AI suite tangled with crypto wallets and token rewards.
Winamp nostalgia for macOS when IINA and Music.app already handle playback.
Enterprise security with 150+ MCP servers, but login-only landing page shows nothing.
This is practical, low-level tooling: the addon runs Opus encode/decode and all RTP I/O on native threads so the Node event loop stays out of the way — exactly what you need when streaming audio to AI models. It bundles helpful primitives (createRtpParameters / createSrtpParameters / createSDP, produceRtp/consumeRtp) but does require native deps (FFmpeg/libopus) and currently only documents macOS/Linux builds.
System audio transcription without API keys—but Whisper Desktop and Otter do this already, better-known.
Wraps OpenClaw deployment, but Telegram bot hosting is a solved, commoditized problem.