databricks-model-serving

You can deploy MLflow models to Databricks Model Serving by using this functionality to upload your model files, log them, and create scalable REST API endpoints for programmatic interaction and predictions.

Yes, you can deploy GenAI agents built with the LangGraph framework to Databricks Model Serving endpoints, enabling programmatic access and interaction with your deployed AI agents.

To deploy and query Databricks Model Serving endpoints, you need to install and configure MLflow, databricks-langchain, langgraph, databricks-agents, and pydantic packages in your environment.

Deploying custom pyfunc models to Databricks Model Serving uses the same deployment workflow as traditional ML models, allowing you to package custom logic into scalable REST API endpoints without special configuration.

You query Databricks Model Serving endpoints by sending programmatic REST API requests to the deployed endpoint, retrieving predictions or other responses generated by your MLflow models or AI agents.

The best way to deploy GenAI agents to Databricks Model Serving is to create the agent file, upload it to your workspace, test it, log the model with MLflow, and finally deploy the agent to the endpoint.