Confluent Adds A2A to Streaming Agents, Launches Multivariate Anomaly Detection

Overview

• Support for the Agent2Agent protocol enters open preview in Streaming Agents to coordinate AI agents using real-time data streams.
• Multivariate Anomaly Detection debuts in early access, analyzing related metrics together to cut false positives and surface issues before they hit applications.
• Streaming Agents pair Anthropic’s Model Context Protocol for context with A2A for inter-agent communication, enabling reuse of outputs and cross-system workflows.
• Confluent routes agent exchanges over Apache Kafka to capture an immutable log, centralize orchestration, and enforce governance, security, and observability.
• Analysts say the additions help businesses act on live signals and differentiate Confluent’s platform, as the company continues toward an unclosed $11 billion IBM acquisition and advances a Real-Time Context Engine in early access.