Everything you need to start building on the MindSpeak AI Baseboard.
Full guides are on the way.
Setup, firmware flashing and detailed pinouts will be published alongside the Dev Kit launch. Follow progress on GitHub in the meantime.
The MindspeakAI Baseboard V1 is a complete voice-assistant peripheral system with a replaceable compute module. It works on its own with cloud AI, and becomes a fully local AI device when a compute module is attached.
For the full hardware story, including an exploded view of each layer, see the Dev Kit page.
The Mindspeak Compute Module (MCM) slot is an open, high-density board-to-board interface. It accepts adapters for Raspberry Pi CM5, NVIDIA Jetson, NPUs or custom AI accelerators.
With a compute module attached, quantized open models such as Llama 3, Mistral and Gemma can run on-device through Ollama, llama.cpp or vLLM. Wake word, far-field audio and inference stay on hardware you own, with no cloud API in the loop.
The baseboard exposes ESP32 JTAG, an XMOS XSYS2/XTAG4 debug header and a UART console. A physical switch controls hardware microphone mute, independent of software.