What is Databricks AI?
Databricks AI, previously marketed as Mosaic AI, is the set of tools on the Databricks platform for building generative AI and machine learning systems. It covers the full path from grounding agents in company data to serving, evaluating and governing them in production. It suits organizations that already keep their data in Databricks and want to build AI applications without moving that data elsewhere.
Main Features
- Build AI agents grounded in enterprise data with Agent Bricks, including synthetic data generation and automated tuning.
- Use models from providers such as OpenAI, Meta and Anthropic, or fine-tune your own.
- Store and search embeddings in a vector database that syncs with source data for RAG.
- Deploy agents, generative AI models and classical ML models through one model serving layer.
- Evaluate agent quality with MLflow, AI-assisted judges and human feedback.
- Control access to models, agents, MCP servers and tools through a gateway with rate limits and guardrails.
Who Should Use It?
- Data and ML engineering teams building AI applications on Databricks.
- Enterprises that need AI agents to answer questions from internal data.
- Teams building retrieval-augmented generation (RAG) apps.
- Platform teams that need governance and guardrails over model and agent access.
- ML teams moving classical models and generative AI into production on one platform.
Pricing
Databricks offers a 14-day free trial with up to $400 in credits.