Expanding MQTT from data movement to coordination across AI agents, connected devices, and enterprise infrastructure
MENLO PARK, Calif., April 2026— EMQ, the company behind the EMQX platform for real-time data, device connectivity, and system coordination across IoT and AI environments, today announced the release of EMQX Enterprise 6.2. Built on MQTT, the platform adds foundational capabilities for agent discovery, coordination, and governance across distributed systems.
With this release, EMQX extends beyond real-time data movement to support event-driven coordination between autonomous agents and connected devices, positioning the platform as a core infrastructure layer for both domains.
EMQX 6.0 established a unified, multi-tenant messaging foundation. EMQX 6.1 extended this with durable streaming and analytics-ready data pipelines. With EMQX 6.2, the platform takes a further step forward, adding new capabilities for agent coordination, governance, and operational control.
“EMQX 6.2 extends MQTT beyond data movement into native agent-to-agent discovery and coordination. By building A2A discovery directly into the broker, we are giving customers a simpler and more practical way to connect agents, devices, and real-time systems on the same infrastructure.”
— Benniu Ji, VP of Product at EMQ Technologies
Enabling Real-Time Discovery and Coordination for Agents
At the center of this release is the introduction of the A2A Registry, a standards-aligned agent discovery system built directly into the MQTT broker.
Agents register themselves by publishing structured “cards” describing their capabilities and metadata to MQTT topics such as $a2a/v1/discovery/{org_id}/{unit_id}/{agent_id}. Other agents subscribe to these topics to discover available peers and receive real-time updates.
Unlike traditional service discovery systems that rely on polling centralized directories, EMQX enables event-driven discovery using retained messages, allowing agents to immediately receive the full set of available peers upon subscription, along with updates as agents join, update, or disconnect.
The registry also provides built-in presence awareness. EMQX automatically annotates discovery messages with live online or offline status based on MQTT session state, allowing consumers to combine discovery and liveness in a single subscription stream.
The A2A Registry follows emerging Agent-to-Agent (A2A) protocol specifications and serves as a foundational capability for upcoming agent-based products, including EMQ’s Device Agent, currently in early access.
In addition, EMQX 6.2 introduces a standardized API specification (/api-spec.md) that enables AI agents to interact with and manage EMQX directly. This allows developers to bootstrap agents from minimal context and perform monitoring, configuration, and operational tasks directly using tools such as Claude Code and Codex without requiring custom integrations, plugins, or specialized skills.
Strengthening Governance and Operational Control at Scale
In addition to agent coordination, EMQX 6.2 introduces several features designed to improve operational reliability, data consistency, and resource control in large-scale deployments.
The new Unified Namespace Governance plugin enforces topic structure and payload schema validation at the broker level. Organizations can define a formal topic model with constraints and JSON schema validation, ensuring that all published data conforms to agreed standards before it reaches downstream systems.
This approach addresses a common challenge in industrial and IoT environments, where topic structures often drift over time. By enforcing rules at the broker, EMQX enables a fail-fast approach that prevents invalid data from propagating through the system.
EMQX 6.2 also introduces subscription-level message filtering, allowing clients to receive only messages that meet specific criteria based on MQTT 5.0 User Properties. This reduces unnecessary data transfer and client-side processing in high-throughput environments.
Dynamic Device Management Without Disruption
To improve device lifecycle management, EMQX 6.2 introduces the ability to dynamically update client keepalive intervals at runtime.
Clients can now adjust their own keepalive settings by publishing to $SETOPTS/mqtt/keepalive, while backend systems can update fleets of devices in bulk using $SETOPTS/mqtt/keepalive-bulk.
This eliminates the need for disconnect-and-reconnect cycles when network conditions or workload patterns change, enabling more resilient operation for large fleets of connected devices.
For example, an electric vehicle manufacturer managing over 100,000 connected vehicles can adjust connectivity behavior across the fleet without interrupting active sessions, reducing power consumption, and improving reliability in constrained environments.
Built for Data, Devices, and Autonomous Systems
EMQX Enterprise 6.2 continues to support its core audiences across IoT, data engineering, and enterprise infrastructure, while expanding to include teams building distributed AI systems.
In addition to established use cases such as data streaming into lakehouse architectures and multi-tenant device management, EMQX now enables:
- Real-time coordination between autonomous agents
- Event-driven discovery and messaging for AI systems
- Unified infrastructure for both devices and agents
This reflects a broader shift in system design, where data producers are increasingly paired with autonomous consumers and decision-makers operating in real time.
