Redis expands Redis Iris with six capabilities for AI agents
Redis announced the expansion of Redis Iris, its real-time context engine for AI agents. The update adds six customizable capabilities that teams can deploy together or separately to deliver the memory, context, and messaging agents require for reliable real-world operation. The capabilities are Redis Intent Router, Redis Agent Playbooks, Redis Agent Memory, Redis Context Gateway, Redis Lens, and Redis Agent Mesh.
Enterprise teams continue to encounter obstacles reaching repeatable production outcomes with agents even though AI models and infrastructure now meet enterprise standards. Agents commonly lack the right context at the right time and in the right form. When they succeed they often do so through inefficient methods such as over-stuffed prompts, repeated LLM calls, and fragile custom integrations that consume more tokens and time than the value created. Rowan Trollope, CEO of Redis, stated that agents need the same foundations as apps and microservices including live app data, runtime state, and service communication, plus new capabilities for memory, context management, and routing. He added that Redis already operates in the hot path as the leading runtime database and that Iris unites these primitives so developers can build agents that function in production without five separate infrastructure products.
Intent Router is generally available and matches the meaning of a request rather than its exact wording before routing it to the appropriate destination, whether a remembered FAQ, an agent skill, a human, or an LLM. The capability also supports agent and model routing, guardrails, and fuzzy if-statements that extend decisions beyond true-or-false logic. Agent Playbooks is generally available and lets teams designate which answers are stable enough to reuse and which require current data plus a fresh execution of a known procedure. Playbooks grow more thorough and accurate with each run, shortening resolution times and lowering token spend.
Agent Memory is generally available and builds memories from user actions as well as documents and other semi-structured data through MemoryETL. It organizes memories in hierarchies for finer control, supplies inspection, audit, and provenance features, and can forget information that should not be retained such as personally identifiable data. Context Gateway is generally available and supplies agents with a single governed entry point to connected context sources. Teams can bring their own MCP servers, cache tools that solve particular problems to reduce token use, and build reusable results that maintain consistency across sessions.
Lens is in beta and makes agent behavior traceable. Every outcome can be linked back to the skill, route, or FAQ that produced it, the memory it drew from, and the originating session, delivering auditable evidence together with visibility into task success, tokens per task, and latency. Agent Mesh is in beta and allows agents to communicate over ephemeral streams while automatically buffering and reconnecting when a connection drops. Durable SSE Streams keep responses flowing through interruptions, and A2A Comms provide a reliable channel for delegation and handoffs.
Iris follows four design principles for real-world agent operation. Context should be navigable rather than siloed. Memory should be adaptive and selective. Context should be efficient to control costs. Context should compound as agents learn from past interactions.
June Lee, CTO of Paradigm, stated that everything in Paradigm is stored in Redis and that it serves as the memory and orchestration layer enabling agents to operate in real time at a scale previously unattainable. Additional customers have placed Redis Iris into production since its initial launch in May 2026.
Redis Iris is available today. Intent Router, Agent Playbooks, Agent Memory, and Context Gateway are generally available. Lens and Agent Mesh are in beta. Redis will host live first-look sessions with the engineering team on November 12 for the Americas and November 17 for EMEA, with registration available at redis.io/resources/videos/redis-iris-full-spectrum.
Harrison Chase, cofounder and CEO of LangChain, observed that memory is core to how agents improve over time and that Redis together with LangSmith's Context Hub gives customers a structured way to manage agent context across environments by connecting live operational data and retrieval into the same system that versions and evolves agent memory. Nick Thompson, Senior Software Developer at Safe in Home, reported that Redis Agent Memory stores and reuses context across coding agents in real time, capturing engineering decisions, bug details, and development context so the information remains available across the team rather than being lost between sessions or tools. A service also reads from and writes to the same memory while monitoring their API, enabling real-time issue detection and bug reporting into their ticketing system.
The State of Context Engineering 2026 survey of IT and AI infrastructure leaders found that nearly all view context as the deciding factor in whether AI works, yet almost none have built the systems to prove it.
