Connect an AI service your organization pays for, so its models can answer your team and your AI members.
Before you start
You need the Manage AI Settings permission, and your organization must have the AI Assistant app switched on. To switch models or capabilities on and off you also need Manage AI Models.
Steps
- Go to Settings → Artificial Intelligence → AI Provider Registry.
- Click Add provider.
- Choose the kind of service: a cloud AI service (OpenAI, Anthropic, Google Gemini, Mistral AI, Cohere, OpenRouter, DeepSeek, xAI, Meta Llama API, Perplexity, Hugging Face, Azure OpenAI), your own server (an OpenAI-compatible endpoint, Ollama, Text Embeddings Inference), or a speech service (Deepgram, ElevenLabs, AssemblyAI, Cartesia, or your own Speaches server). Any other service that speaks the OpenAI API (Groq, Together AI, Fireworks AI, vLLM, LM Studio and many more) is added as an OpenAI-compatible endpoint.
- Give it a name, and paste its API key (a server of your own may not need one). Your own servers (Speaches included) and Azure OpenAI also need a Base URL; for the other services you can leave it empty.
- Click Test connection. Kamo makes one small request and tells you whether the service answered, with a few of the models it lists.
- Click Save and discover models. Kamo saves the provider and lists the service's models straight away.
After you add it
- Models — models Kamo recognizes arrive switched on; anything it cannot identify arrives switched off so you can check it first. Switch each on or off in the provider's Models room.
- Capabilities — open a model to see what it can do (text, vision, tool calling, embeddings, speech and more), where each fact came from, and whether it is Verified.
- Verify — click Verify on a capability, or Verify enabled models on the provider, and Kamo runs a small real request to prove it. AI team members only use capabilities that are verified; people can use anything that is switched on. Verify a model before you count on it for an AI member: until then, the AI member skips it.
- Prices — a model's Pricing room holds what it costs. Kamo fills it in for models it knows; set it yourself for the others, or their usage shows no cost.
- Order — when nothing more specific is asked for, Kamo tries your providers from the top of the list. Use the arrows on the list to change the order. To prefer particular models for a kind of work, use Routing & Budgets; see Routing & Budgets Explained.
If the key changes
Paste the new key in the provider's Connection room and save. If you point the Base URL at a different server, Kamo asks for the key again, so a stored key is never sent to an address it was not entered for.
Related articles
Other guides that answer questions close to this one.
AI Providers Explained
An AI provider is your organization's account with an AI service: where it is, and the key that pays for it. The AI Provider Registry (Settings → Artificial Intelligence → AI Provider Registry) keeps every provider you…
Routing & Budgets Explained
Two questions every AI request answers: which model does the work, and may this AI member spend any more today? Settings → Artificial Intelligence → Routing & Budgets is where you answer both for your organization. You…
How to Add an MCP Server
Connect an MCP server so the AI can use its tools. See MCP Servers Explained first. Before you start The Manage AI Settings right, and the server's MCP endpoint URL (for example https://tools.example.com/mcp). Servers…
How to Review AI Usage and Costs
See how much your organization is using AI, broken down by provider, model, person and day. Before you start You need the Manage AI Settings permission. Steps 1. Go to Settings → Artificial Intelligence → Usage &…
