Confluent Tackles the Data Problem Blocking Real AI

Eight in ten companies say data limitations are holding back their AI ambitions. Confluent, now part of IBM, thinks it has the fix.

Confluent, the data streaming company acquired by IBM, has announced a set of new capabilities designed to address two problems that routinely prevent AI projects from reaching production: fragmented developer tooling and security teams unwilling to expose sensitive data to AI pipelines.

The release spans four areas. First, a fully managed Model Context Protocol server and a feature called Agent Skills allow AI to build, manage, and debug streaming operations using natural language, encoding best practices so those operations run consistently against organizational standards. 

Second, a new built-in machine learning function for Apache Flink detects and redacts personally identifiable information directly in Flink SQL, without custom code or external services — a capability aimed squarely at regulated industries such as financial services, healthcare, and insurance. 

Third, support for Azure Private Link ensures that AI workloads connect to external models and services over Microsoft’s private network backbone rather than the public internet. Fourth, a free open-source dbt adapter brings Flink SQL into the data pipeline framework that most data engineers already use, lowering the barrier to real-time data processing for teams that would otherwise need to learn new tooling.

The company also added support for TimesFM models for anomaly detection, alongside Anthropic and Fireworks AI models that developers can call directly within Flink stream processing workflows.

Also Read: 100 Things Google Announced at I/O 2026

The announcements extend recent integrations unveiled at IBM Think, where Confluent Cloud was woven further into IBM’s watsonx.data platform to provide a real-time context layer for AI across hybrid environments.

“Most AI projects fail before they reach a single customer because the data layer breaks down,” said Sean Falconer, Confluent’s head of AI. “Teams have the models and the mandate, but security risks and fragmented data stop them from shipping.”

The challenge is well documented. A McKinsey report found that eight in ten companies cite data limitations as a primary obstacle to scaling AI agents — a figure that points to infrastructure gaps rather than a shortage of ambition or investment.