We are entering a new phase in how enterprises interact with data.

For years, data platforms have been built with humans in mind, dashboards, SQL editors, APIs. That model is now being challenged by a new interface: AI agents. These agents don’t just query data, they reason, orchestrate workflows, and take action.

With this release, watsonx.data moves decisively in that direction.

We are introducing a managed remote MCP server, further positioning watsonx.data as an agent-ready data platform.

From APIs to agents

Traditional APIs were designed for developers. Each integration required custom logic, tight coupling, and manual orchestration.

Agents operate differently. They require:

• Semantic discovery of capabilities
• Standardized interfaces
• The ability to reason about available actions

This is where MCP comes in. With MCP, watsonx.data exposes its capabilities as structured, discoverable tools, enabling agents to dynamically understand what they can do and execute actions in a governed way.

What’s new in watsonx.data 2.3.2

This release introduces a platform-managed MCP server for watsonx.data SaaS. Instead of deploying and managing a local MCP server, users can now connect directly to a remote, fully managed endpoint built for enterprise use.

Key capabilities include:

• Engine management: Full lifecycle management of Presto and Spark engines
• Catalog operations and data discovery: Explore catalogs, schemas, and tables
• Query execution and optimization: Execute SQL and use EXPLAIN / EXPLAIN ANALYZE
• Spark application management: Submit and monitor Spark jobs
• Data ingestion: Load from S3 and IBM Cloud Object Storage (CSV, Parquet, JSON)
• Secure access via enterprise IAM: Token-based authentication with governance controls

Built for enterprise environments

MCP in watsonx.data is not just about access, it is about governed access.

All operations executed through MCP are governed by watsonx.data core services, ensuring that access control, policies, and auditing remain consistent with platform enforcement.

This means agents operate within the same security and compliance boundaries as human users.

What this enables

By exposing data as tools, watsonx.data enables a new class of use cases:

• AI agents that autonomously explore and analyze enterprise data
• Intelligent pipelines that adapt based on data conditions
• Multi-step workflows combining query, ingestion, and operations
• Seamless integration with agent frameworks and enterprise AI platforms

This is not just a new feature. It is a new way to interact with data.

See it in Action

What’s next

The shift to agent-native data access has started. And watsonx.data is ready for it. New tools and capabilities will continue to be added.

To explore all available MCP tools,  visit this link