Repository navigation
Added
- Settings is split into General, Users, and AI pages, each with its own address you can open or share directly. (#23)
- You add each AI server once and then choose which one fills each job: chat, speech to text, text to speech, and embeddings. (#23)
- Bloud reads the model list from an AI server by itself when you save it and when you open the page, so you pick a model from a list instead of typing its name. (#23)
- Hermes can transcribe a spoken message and speak its reply. Point the Speech to text and Text to speech rows in Settings → AI at a server, then turn voice replies on in the chat. (#24)
Fixed
- An AI server that cannot be reached is still saved, and the page now tells you it could not reach it instead of discarding what you typed. (#23)
- An AI server that does not publish a model list now lets you type the model name. (#23)
- The developer graph no longer shows an AI server that no job uses. (#23)
Install:
curl -fsSL https://github.com/bloud-os/bloud/releases/latest/download/install.sh | sudo shThe release carries SHA256SUMS, and install.sh checks the .deb against it.