Confluent Unveils Streaming Agents for Scalable Real-Time Agentic AI

Confluent has launched Streaming Agents in Confluent Cloud, a new feature for Apache Flink that simplifies the integration and scalability of AI agents for real-time data processing. This aims to address challenges in moving from prototype to production-ready AI, streamlining workflows and accelerating the adoption of agentic AI. The feature allows agents to invoke external tools and connect securely to various systems for enhanced decision-making capabilities.

Confluent, Inc. (Nasdaq: CFLT) has introduced Streaming Agents, a new feature in Confluent Cloud for Apache Flink that aims to address the challenges of moving from prototype to production-ready AI. This innovative capability simplifies the integration and scalability of AI agents for real-time data processing, thereby streamlining workflows and accelerating the adoption of agentic AI.

Streaming Agents unify data processing and AI workflows, providing easy and secure connections to various business systems, including large language models (LLMs) and embedding models. This unification facilitates the seamless integration of real-time data, enabling AI agents to make informed decisions and act accordingly.

The new feature addresses the common issue of organizations being stuck in "prototype purgatory," where AI initiatives fail to reach production due to complex workflows and data integration challenges [1]. According to IDC research, while enterprises run an average of 23 generative AI proofs of concept between 2023 and 2024, only three reach production, and just 62% of these meet expectations [1].

Streaming Agents bring agentic AI directly into stream processing pipelines, allowing teams to build, deploy, and orchestrate event-driven agents using Apache Kafka and Apache Flink. This integration ensures that agents have access to fresh contextual data from real-time sources, enabling them to quickly adapt and communicate with other agents and systems as conditions change.

Key features of Streaming Agents include:
- Tool calling for context-aware automation: Agents can invoke external tools such as databases, SaaS platforms, or APIs using Model Context Protocol (MCP), allowing them to take meaningful action based on business context.
- Secure connections: Streaming Agents enable secure connections to models, vector databases, and other systems, protecting sensitive credentials and encouraging reuse across multiple tables, models, and functions.
- Enhanced AI accuracy: By enriching streaming data with non-Kafka data sources, such as relational databases and REST APIs, Streaming Agents improve the accuracy of AI decision-making, vector search, and retrieval-augmented generation (RAG) applications.
- Replayability for iteration and safety: Agents can be developed and evaluated using real data without live side effects, enabling dark launches, A/B testing, and faster iteration.

Confluent’s Chief Product Officer, Shaun Clowes, emphasized the importance of fresh business context for AI agents, stating, “Even your smartest AI agents are flying blind if they don’t have fresh business context. Streaming Agents simplifies the messy work of integrating the tools and data that create real intelligence, giving organizations a solid foundation to deploy AI agents that drive meaningful change across the business” [1].

Stewart Bond, Vice President of Data Intelligence and Integration Software at IDC, underscored the need for agentic AI solutions that offer easy, secure integration and leverage real-time data for intelligent action [1].

Streaming Agents are available today in open preview, and additional features such as Private Networking Interface (PNI) support for Enterprise clusters, Tableflow’s new Upsert feature, and Custom Single Message Transformations (SMTs) for fully managed connectors are also part of the Confluent Cloud updates [1].

References:
[1] https://finance.yahoo.com/news/confluent-unlocks-scalable-real-time-130000395.html
[2] https://www.stocktitan.net/news/CFLT/confluent-unlocks-scalable-real-time-agentic-ai-with-streaming-vkszn0qt7kzs.html