Dodaios is an open-source agent OS for your desktop: it runs your coding agents, browser and apps by voice or chat, on your own machine. Formerly Personal Jarvis.
Python agent-orchestrationai-agentai-coding-assistantclaude-codecoding-agentcomputer-usedesktop-applocal-firstmcppersonal-jarvisvoice-assistantvoice-coding
docs: cover three local-model first-run problems (#895)
ANO3457eb7a02ef0344
diff --git a/docs/product/personalize-and-connect/local-ai-providers.md b/docs/product/personalize-and-connect/local-ai-providers.mdindex ce57f40643..94f022b4be 100644--- a/docs/product/personalize-and-connect/local-ai-providers.md+++ b/docs/product/personalize-and-connect/local-ai-providers.md@@ -167,6 +167,9 @@ tools are local.|---|---|---|| Engine installed, model not ready | Missing or incomplete model files | Keep the app open, check space and the error, then select **Try again**. || Ollama test reports no model | Ollama is running with no downloaded model | Download one under **Models on this machine** on the Ollama card, then retest. |+| Ollama card says "dictation polish: No answer within 8s" | The optional wording test uses a separate model (`qwen3.5:4b` by default), not the chat model | Run `ollama pull qwen3.5:4b`, or choose an installed dictation polish model, then test again. |+| Whisper downloads, but readiness reports missing files | Older readiness checks also expected repository metadata | Update to a release containing [the readiness fix](https://github.com/Dodaios/Dodaios/pull/841). On older versions, use [the full-snapshot workaround](https://github.com/Dodaios/Dodaios/issues/835) in Dodaios's Python environment, then reopen the app. |+| Pipeline voice with Ollama says "no API key", although chat works | No Ollama model is selected on an older version | Select an installed model on the Ollama card and start a new voice session. This was fixed by [local model discovery](https://github.com/Dodaios/Dodaios/pull/838). || Self-hosted server is unreachable | Stopped server, wrong address, or blocked network | Restart it, copy its address, select **Save**, and retest. || Chat works, but an action or image fails | The model lacks tools or vision | Choose a capable Ollama model or another suitable provider. || Replies are very slow | The model is too large for available memory or compute | Choose a smaller Brain, use Nemotron for faster speech, or move the server. |