Confluent’s Real-time Data Broker Powers AI Agents

By Betsy Burton

October 29, 2025, Confluent announced the launch of Confluent Intelligence, a suite of products designed to help enable real-time AI. The suite includes Real-Time Context Engine and enhanced Streaming Agents.

Confluent states that it has developed these services to help organizations tackle the challenge of supplying current, governed, and accurate data to LLMs and AI agents, positioning the firm as a central orchestrator in the burgeoning agentic AI stack.

What Was Announced

Confluent’s announcement centered on two technologies that leverage Confluent’s Cloud platform leveraging Apache Kafka and Apache Flink.

Confluent Real-Time Context Engine:

The Real-Time Context Engine is a new, managed service. It is designed to act as a singular gateway for real-time structured data, using MCP, to serve real-time context to any external AI application, LLM or copilot.

Its primary capability is to unify the data processing, reprocessing, and serving layers, ensuring that the AI’s knowledge base is current and contextually relevant. This service provides a snapshot of the business’s state derived from continuous data streams.

Confluent Streaming Agents:

Confluent’s existing product Streaming Agents has been enhanced to not only use Flink to embed logic directly into data, but also use Flink to build, deploy, and orchestrate event-driven AI agents natively on Confluent Cloud. This means agents will be able to observe, decide, and act on real-time events as they occur within the data stream itself.

Enhancements also include native support for RAG and MCP to simplifying the process of creating intelligent automation that is grounded in fresh data.

Strengths

Confluent’s new offerings are built on a powerful foundation, unifying real-time AI and data processing. By embedding Apache Flink-powered Streaming Agents directly within event streams, they create a single, efficient environment for data transformation, AI reasoning, and action, which significantly reduces latency and simplified integration.

A key advantage lies in the management, auditability, and logging features integrated into the Real-Time Context Engine. This addresses crucial business demands for explainable, trustworthy, and compliant AI, making it particularly appealing for regulated sectors.

Economically, these innovations promise increased efficiency and faster time-to-value by streamlining the process of contextualizing data for AI, thus lowering operational costs and accelerating development cycles for real-time AI applications.

Challenges

Despite their promise, these technologies face certain hurdles. Both the Real-Time Context Engine and Streaming Agents are not yet generally available, meaning early adopters might encounter platform immaturity and evolving features.

Another challenge is the complexity of managing sophisticated Apache Flink environments, even within a managed service.

While Confluent is based on open technologies and APIs, its suite is more of a broker/gateway than an open platform. The Real-Time Context Engine itself is a proprietary, fully managed service offered exclusively on Confluent Cloud. This could easily lead to increased vendor lock-in and associated costs. Clients must carefully weigh the operational benefits against long-term dependencies.

Furthermore, Confluent operates in a highly competitive market, facing strong rivals such as Databricks and Snowflake, as well as hyperscalers like AWS, Azure, and GCP, all of whom are rapidly advancing their own integrated streaming and AI capabilities. In addition to increased competition, it very well could be a acquisition candidate.

Bottom Line

Confluent has put forward a compelling technology in its real-time data broker. It will help increase that availability of real-time data for AI systems, and particularly AI agents. However, organizations should be very cautious about the data quality common governance and cleansing that’s needed to support quality AI systems and agents.

It is important to remember that “garbage in” just leads to “garbage out.” While Confluence technology may provide the connection between AI systems and real time and some of the management capabilities, it does not solve the business data management and governance challenges organizations will increasingly be faced with.