Redis expands Iris with six capabilities for production AI agents

Six capabilities in one runtime

Redis Iris now combines six customizable capabilities that teams can use together, independently or in any combination. Redis says the capabilities are designed to give agents the memory, context, routing, communication and live application data they need without requiring teams to connect multiple infrastructure products.

• Intent Router, available generally, matches the meaning of a request and sends it to a remembered FAQ, an agent skill, a human or an LLM. It also supports agent and model routing, guardrails and decisions that go beyond simple true or false conditions.
• Agent Playbooks, also generally available, distinguish stable answers that can be reused from tasks requiring current data and a fresh procedure. Redis says playbooks become more thorough and accurate as they run.
• Agent Memory builds memories from user actions, documents and other semi-structured data through MemoryETL. It provides hierarchical organization, inspection, audit and provenance, and can forget information such as personally identifiable information.
• Context Gateway provides a governed entry point for connected context sources. It supports customer-provided MCP servers, tool caching and reusable results across sessions.
• Lens, in beta, traces outcomes to the skill, route or FAQ involved, the memory used and the source session. It also provides visibility into task success, tokens per task and latency.
• Agent Mesh, also in beta, supports communication over ephemeral streams, automatic buffering and reconnection. Durable SSE Streams keep responses flowing through interruptions, while A2A Comms support delegation and handoffs.

Designed for agents that operate in the real world

Redis frames Iris as a layer between an agent and the data and tools it needs to act. The company says production agents must handle unpredictable human behavior, multiple channels and handoffs, each of which can create failure points.

The platform is built around four principles: navigable context that connects data and tools, adaptive memory that knows what to remember and forget, efficient retrieval that limits unnecessary costs, and compounding context that learns and grows over time. Redis says these principles help agents find relevant information, retain what matters, reduce operating costs and improve through interactions.

According to Redis, Iris can also reduce custom integration work across retrieval, memory, routing and messaging because it runs on Redis technology. The company says the result is more outcomes per token, although the announcement does not provide performance figures.

Iris moves from launch to production use

Iris first launched in May 2026, and Redis says customers have since used it in production. June Lee, CTO of Paradigm, said: “Everything you see in Paradigm is stored in Redis. It is the memory and orchestration layer that lets our agents work in real time, at a scale that wasn't possible before.”

For users, the announcement positions Iris as infrastructure for operating agents after they have been built, with features for explainability, memory control, routing and communication. Redis invites teams to start with one business-critical use case and offers 5,000 free messages. The company also lists a technical deep dive, a demo video and webinars, including live sessions on Nov. 12 for the Americas and Nov. 17 for EMEA.