Local-first AI chat with Ollama and optional CLI providers
Started Last update
Wollama runs on the web, on Electron desktop and on Capacitor mobile. It uses Ollama by default for local inference, voice and retrieval features. For text chat, its server can also call installed Codex, Claude Code or Kimi Code CLIs using their existing sign-ins. Conversations and companions remain accessible offline and sync through RxDB and PouchDB.
flowchart TD P["idea / needs"] --> S["wollama"] S --> A["Model provider registry"] A --> O["Ollama · default local inference"] A --> L["Codex · Claude Code · Kimi Code CLI text chat"] S --> B["web + Electron + Capacitor"] S --> C["offline RxDB <-> PouchDB sync"] S --> D["voice · RAG · skills · hooks · web agents"]
Wollama runs on the web, on desktop through Electron and on mobile through Capacitor. Ollama is the default for local inference, and installed Codex, Claude Code and Kimi Code CLIs can handle text chat through their own services. Conversations, companions and settings stay available offline and sync through RxDB and PouchDB when the server returns.