Blog
Talks and guides for a local-first AI workspace
Notes from BKK Hack, how Velociti keeps the project on your computer, how you switch models, and how local GGUF and your keys sit in one desk.
BKK Hack local AI
Beyond the Clouds. Notes from BKK Hack
Persistent workers, open-weight models, and a local-first workspace that keeps prompts, context, and responses on your computer. Download the BKK Hack slides.
local-first AI workspace
What a local-first AI workspace is
A local-first AI workspace runs compatible models on your computer, keeps the project with you, and sends a turn to a supported provider only when you choose.
switch AI models keep project
Switch AI models without losing the project
Change from a local model to ChatGPT, Claude, Grok, or another supported provider without reloading files, instructions, or approved memory.
AI that stays on your computer
AI work that stays on your computer
Prompts, context, responses, and thinking stay on your machine with the project. They are not used as training data.
run local LLM on your computer
Run local models on your computer
Install a compatible local model in Velociti. GGUF weights can live in system RAM, including DDR5, as well as GPU VRAM. Placement follows measured headroom.
Velociti Free vs Plus
Velociti Free vs Plus
Free is a complete personal AI workspace. Plus adds continuity: shared work, Hermes Bots, schedules, and recovery.
Also see the models Velociti works with and honest comparisons.