Rixoffice is provider-agnostic: one Settings screen decides whose intelligence powers the AI composer in all five apps.
Choose a provider
Open Settings, then the AI provider section. You will see these options:
| Provider | What you need | Best for |
|---|---|---|
| Built-in Rix AI | A Rixcart sign-in and an active plan | Zero setup, metered credits |
| OpenRouter | An OpenRouter API key | One key for many models |
| Anthropic (Claude) | A Claude API key | Long-document reasoning |
| OpenAI | An OpenAI API key | GPT models with your own billing |
| Google Gemini | A Gemini API key | Fast multimodal work |
| DeepSeek | A DeepSeek API key | Cost-effective reasoning |
| Ollama | Ollama installed locally | 100% offline, fully private AI |
| Custom | An OpenAI-compatible base URL | Self-hosted or niche providers |
Use the built-in Rix AI
- Sign in with your Rixcart account.
- Make sure a plan is active — Starter includes 20 credits, Standard AI 99, Pro 399 and Premium 1,200 per month.
- That is it: open any document and the composer works immediately. Credits appear in the app and on the credit usage page.
Connect your own key
- Choose the provider, for example OpenRouter.
- Paste your API key. The key is stored locally on your device.
- Pick the model you want from the model list.
- Press Test — Rixoffice sends a tiny request to verify the key. A green result means you are ready.
BYO keys are never metered: no credits are consumed when you use your own provider.
Run AI fully offline with Ollama
- Install Ollama from ollama.com on Windows, macOS or Linux.
- In a terminal, pull a model, for example: ollama pull llama3
- Keep Ollama running (it listens on localhost by default).
- In Rixoffice Settings choose Ollama, then select the model you pulled.
- Try the composer — it now works with no internet connection at all.
Ollama is available on the Pro API & Local plan and above. This is the strongest privacy option: documents, prompts and answers never leave your machine.
Use a custom endpoint
Self-hosting an OpenAI-compatible server such as llama.cpp server, LM Studio or vLLM? Choose Custom, enter the base URL (for example http://localhost:8080/v1), add an API key if your server requires one, and test.
Switching providers any time
You can change provider per moment, not per month: use built-in Rix AI on a laptop trip, Ollama on a flight, and your Claude key for heavy writing — the composer, agents and per-app skills behave identically regardless of the engine behind them.