AI Providers and BYOK
Connect your own AI providers, understand provider routing, and choose local or cloud models.
MachinesFluent can process clipboard text, images, screenshots, and spoken instructions with the AI provider you choose.
BYOK means "bring your own key." Instead of being locked to one bundled AI account, you connect your own provider account or local model runtime where supported.
What BYOK Means
When you use a cloud AI provider:
- you provide the API key or provider connection
- requests are billed by that provider according to your provider account
- provider-side data handling follows that provider's terms, settings, and retention controls
- MachinesFluent sends the content needed for the workflow you triggered
When you use a local provider such as Ollama or LM Studio, the model can run on your own machine or local network depending on your setup.
Supported Provider Set
The current product context lists these AI providers:
| Provider | Typical role |
|---|---|
| OpenAI | General AI processing, writing, coding, vision-capable workflows depending on model |
| Anthropic | Long-form reasoning, writing, coding, analysis |
| Google Gemini | General AI processing and multimodal workflows depending on model |
| DeepSeek | Coding and reasoning workflows |
| OpenRouter | Access to multiple hosted models behind one provider account |
| Cerebras | Fast hosted inference workflows |
| Groq | Fast hosted inference workflows |
| Mistral | Hosted language models |
| Perplexity | Web-grounded answers and research-style responses |
| Alibaba | Hosted language models |
| Zhipu | Hosted language models |
| Replicate | Image-generation or model-hosting workflows depending on configured model |
| Ollama | Local or self-hosted model workflows |
| LM Studio | Local model workflows |
Provider availability, model lists, pricing, and data policies can change. Check the provider before sending sensitive material.
Safe Setup Pattern
Start with a low-risk prompt
Test with a harmless sentence, not private customer data or credentials.
Confirm the selected provider
Make sure the active provider and model are the ones you intended to use.
Add your API key or local endpoint
Use the provider settings inside MachinesFluent. Do not paste secrets into prompts, notes, or support messages.
Test output before enabling auto-paste
Preview the result first, then enable faster paste behavior once the prompt is reliable.
What Gets Sent To The Provider
The provider receives the content needed for the action you trigger. Depending on the workflow, that can include:
- clipboard text
- selected text
- copied images or screenshots
- spoken instructions converted to text
- system prompt or saved prompt content
- optional context files you have configured