Confluent adds support for the Agent2Agent protocol, using Apache Kafka to orchestrate inter-agent communication among streaming agents and for anomaly detection.

The New Stack is a publication covering trends and technologies in cloud-native development, DevOps, and software delivery. Developers can learn about containerization, Kubernetes, and cloud computing, as well as explore topics such as microservices architecture, serverless computing, and continuous integration/continuous delivery (CI/CD) pipelines.

The New Stack

Confluent has announced several major platform updates including support for the Agent2Agent (A2A) protocol, multivariate anomaly detection, and Queues for Kafka (KIP-932). The A2A integration uses Apache Kafka to orchestrate asynchronous inter-agent communication, enabling agents built outside Confluent to become event-driven. Every agent decision is logged in real-time to system tables (Kafka topics with structured schemas), providing auditability and traceability. The multivariate anomaly detection uses lightweight statistical ML techniques like ARIMA and MAD that learn in-stream without batch training, making them cost-effective for IoT and operational monitoring. Queues for Kafka extends Kafka's traditional pub/sub model to support message queue semantics within the same topic, reducing infrastructure complexity for developers needing both paradigms.

Confluent adds A2A support, anomaly detection, and Queues for Kafka in major platform update

How Kafka powers asynchronous A2A communication

System tables give agents a built-in audit trail

Multivariate anomaly detection runs in the stream, not in a batch

Queues for Kafka: pub/sub and message queues in a single topic