Build, deploy, and orchestrate event-driven agents natively on Apache Flink® to power intelligent real-time automation across your organization.
Streaming Agents enable you to build, test, deploy, and orchestrate event-driven agents using fully managed Flink and Kafka on a unified platform. With access to real-time contextualized data, they’re uniquely suited for enterprise workflows – monitoring and taking instant, informed action the moment operational events occur.
Whether you’re automating fraud prevention or personalizing product recommendations, Streaming Agents help you turn agents from brittle demos into production-ready, event-driven multi-agent systems.
Watch Demo
Simplify development by embedding agents in the data stream and directly working with AI models in Flink. Combine stream processing jobs (e.g., join, filter, enrich) and agentic workflows in the same pipeline, reducing operational complexity while empowering every engineer to be an AI engineer using familiar Flink APIs.
Enterprises need agents to be the eyes and ears for the business in motion, but most agents lack real-time awareness. Running as Flink jobs in stream processing pipelines, Streaming Agents can access the latest, most accurate view of the business, at the moment a decision needs to be made – zero data polling or delay. They’re always on, state-aware, and retain context over time.
Integrate without stitching together disparate systems. Use your preferred tools and Flink APIs with native support for connecting to large language models (LLMs) and embedding models, tool calling with Model Context Protocol (MCP), and data enrichment from other systems with External Tables and Vector Search.
Maintain full visibility and control with secure connections, RBAC, built-in replayability, dark mode testing, and governed, traceable event flows.
Streaming Agents ingest live customer interaction data, leverage context from internal systems (e.g., purchase history, clickstream, inventory), use an LLM to understand preferences and intent, and then instantly tailor product recommendations–continuously adapting to maximize engagement and conversion.
By monitoring operational data streams–e.g., call drop rates, signal quality, bandwidth usage–Streaming Agents detect unusual patterns, correlate signals across data sources, use an LLM to assess likely root causes and impact, then trigger remediation steps or escalate to the right teams–enabling faster mean time to resolution (MTTR).
Streaming Agents automatically trigger MCP tools to scrape competitor websites for real-time pricing on ordered products, update the customer’s order to match lower prices, notify the customer of their savings via email, and log each price match in a CRM system.
“Good AI needs good data. Confluent is our trusted source of truth… the data streaming platform provides context and orchestration for our AI agents to automate workflows and accelerate our smart city transformation.”
Ready to build your first agent? Sign up and receive $400 to spend during your first 30 days.
Use Streaming Agents on Confluent Cloud to make production-ready, event-driven multi-agent systems a reality.