IBM watsonx.ai
IBM Cloud · AI
IBM's AI studio: Granite and other models for text, chat, embeddings and deployments
The client pastes their API key from IBM Cloud's settings. We store it encrypted.
Tested against IBM watsonx.ai's live API.
Building with IBM watsonx.ai? The docs have setup steps and code for every tool.
Open the IBM Cloud docsTools
Your agent calls any endpoint of IBM watsonx.ai's API with the request tool, with the client's access attached.
Base URL for request
https://{region}.ml.cloud.ibm.com/ml/v1
request
HTTPS ANY /v1/clients/{clientId}/watsonx-ai/{path}
Any endpoint of IBM watsonx.ai's own API, with the client's access attached.
- servicestringrequired
- This service: watsonx-ai
- methodstringrequired
- GET, POST, PUT, PATCH or DELETE
- pathstringrequired
- The path under https://{region}.ml.cloud.ibm.com/ml/v1
- queryobjectoptional
- Query parameters
- bodyobjectoptional
- A JSON body
Over MCP every tool also takes client_id, from list_clients. Over HTTPS send it as clientId to https://api.operate.as/v1.
What the client approves
- Use the IBM Cloud key you paste, with only the access you gave that key
Get started
Send your next client one link
Add a client, select the services, copy the link. You'll get an email the moment their access is ready. Start free, no card required. Paid plans are one flat monthly price, and a busy month never takes your agents offline.