Confluent launches advanced Streaming Agents to boost AI agility

Confluent has announced new advancements to its Streaming Agents offering, aimed at simplifying the development and scaling of event-driven artificial intelligence (AI) agents.

The enhancements to Streaming Agents include the introduction of Agent Definition, which allows teams to create production-ready AI agents using only a few lines of code. The company states that the features seek to reduce coding time, increase observability for debugging and testing AI systems, and provide enterprise-level governance controls. These additions also feature a Real-Time Context Engine intended to supply fresh context and governance for deploying trustworthy AI agents.

Many organisations looking to adopt agentic AI solutions face complexities related to data integration, troubleshooting failures, and the challenge of scaling systems beyond monolithic architectures. For businesses with high operational stakes, real-time responsiveness is often required, but existing AI frequently cannot act on critical events autonomously.

Commenting on the criticality of real-time data processing, the IDC MarketScape: Worldwide Data Platform Software 2025 Vendor Assessment states, "With advancements in agentic AI and data processing automation, real-time data processing and streaming analytics capabilities are essential. The ability to process data as it enters the system is crucial for time-sensitive applications such as fraud detection, personalization, and operational monitoring, where delays can result in lost opportunities or increased risks."

Streaming Agents incorporates the event-streaming and processing capabilities of Apache Flink to offer large-scale, low-latency, and fault-tolerant workflows. By integrating agent functionalities-including large language models, memory, and orchestration-directly into event streams, users can monitor business operations with up-to-date data. This enables AI agents to observe, make decisions, and act without the need to piece together separate systems.

"Today, most enterprise AI systems can't respond automatically to important events happening in a business without someone prompting them first," said Sean Falconer, Head of AI at Confluent. "This leads to lost revenue, unhappy customers, or added risk when a payment fails or a network malfunctions. Streaming Agents brings real-time data and agent reasoning together so teams can quickly launch AI agents that observe and act in real time with the freshest, most accurate data."

The recent updates to Streaming Agents, now available in Open Preview, allow organisations to deploy and orchestrate event-driven AI agents on Confluent Cloud. This aims to unify data processing with AI workflows, thereby accelerating the development of agentic AI within enterprises.

The three core new features introduced are:

• Agent Definition: Allows teams to codify agent workflows with minimal code, facilitating reuse and easy optimisation. Agents can be programmed for complex tasks using iterative tool calling.
• Observability and Debugging: The platform offers expanded visibility into agent activity, making it easier to test, approve, and oversee AI operations. Developers gain the ability to trace, replay, and compare interactions, supporting faster iteration and a tamper-proof log for compliance.
• Real-Time Context Engine: This managed service provides updated context for AI agents, supporting reprocessing of contextual data and adherence to security and compliance requirements. It includes authentication, role-based access control, and audit logging.

Streaming Agents can operate across major cloud platforms, including Amazon Web Services, Microsoft Azure, and Google Cloud, and can communicate with external agents built on frameworks such as LlamaIndex. The system is compatible with event-driven workflows touching technologies such as Clickhouse, MongoDB, and Snowflake. According to Confluent, customers can leverage system integrators and experts, including Infosys, to assist with the deployment and scaling of Streaming Agents.

An open-source variant named Flink Agents is also available as part of the Apache Flink project. Flink Agents is a result of collaboration between Alibaba Cloud, Confluent, LinkedIn, and Ververica, offering a framework for building long-running, event-driven AI agents directly within Flink's runtime.