San Francisco, November 4, 2026 /PRNewswire/ -- Redis today announces the launch of the next phase of features for Redis Iris. Redis Iris is its real-time context engine for AI agents. Redis Iris now offers six customizable capabilities that can be used in combination or independently, providing agents with the memory, context, and messaging capabilities necessary for reliable operation in the real world: Redis Intent Router, Redis Agent Playbooks, Redis Agent Memory, Redis Context Gateway, Redis Lens, and Redis Agent Mesh.

Redis indicates that the AI models and infrastructure are now sufficient to meet corporate needs, but few teams have achieved repeatable, production-level success in proxying. Proxies often fail to remember the correct context in the right way at the right time. Even when some proxies can create value, they are often inefficient, suffering from issues such as excessive use of prompts, repeated calls to large language models, and fragile custom integrations, which results in higher costs in token and over time, these costs outweigh the value they provide.

Redis CEO Rowan Trollope stated: "Proxies require the same foundation as applications and microservices: real-time application data, runtime state, and service communication. They also need new capabilities such as memory, context management, and routing. As a leading runtime database for applications, Redis is already on the critical path. Developers should not have to rely on five different infrastructure products just to make proxies work. Iris integrates these fundamental capabilities into one place, allowing developers to build proxies that can run in a production environment."

Redis indicates that Iris has provided context, memory, and messaging capabilities for production environments from the very beginning, while also allowing teams to customize each of these capabilities for their respective agents. This release includes:

• Intent Router (officially available, GA)It matches based on the meaning of the request rather than its wording, and sends it to the correct destination, whether that’s a memorized FAQ, an agent skill, a human operator, or a large language model. It also supports routing between agents and models, guardrail mechanisms, as well as “fuzzy” if statements, allowing decisions to be no longer limited to just true or false.
• Agent Playbooks (officially available, GA)Let the team define which answers are stable and should be reused, and which answers require current data as well as a new round of execution of the known processes. Redis indicates that as the number of runs increases, Playbooks will become more comprehensive and accurate, thereby reducing problem-solving time and lowering the token overhead.
• Agent Memory (officially available, GA)Build memories from user behavior, documents, and other semi-structured data through MemoryETL, and organize them in a hierarchical structure to achieve finer-grained control. At the same time, it provides functions for checking, auditing, and tracing the source. This feature can also erase information that should not be retained, such as personal identification information.
• Context Gateway (officially available, GA)Provide a unified and governed entry point for proxies to access connected context sources. Teams can connect to their own MCP servers, cache which tools can solve specific problems to reduce the use of token, and build reusable results to maintain consistency across different sessions.
• Lens (Test version, beta)Makes proxy behavior traceable. Redis indicates that each result can be traced back to the skill, route, or frequently asked question that provided that result, to the memory upon which it was learned, and to the source session. This provides auditable evidence while making visible the task success rate, the token consumption for each task, and the delay.
• Agent Mesh (Test version, beta)Allows agents to communicate via temporary streams, and automatically buffers and reconnects when the connection is interrupted. SSE Streams ensures continuous transmission during interruptions, while A2A Comms provides a reliable channel for delegation and handover between agents.

Redis indicates that by combining these capabilities, it helps ensure that the agent remembers the correct information, answers in the right way, remains transparent in each operation, and delivers more results with less token. Since Iris runs on the Redis technology that the team is already familiar with, enterprises can reduce the custom integration work typically required for connecting retrieval, memory, routing, and message passing.

Redis indicates that production-level proxies must handle unpredictable human input, multiple channels, as well as handovers between proxies, and any of these links can become a point of failure. Iris is built around four principles: the context should be navigable rather than isolated from each other; memory should be adaptive and selective; the context should be efficient to control costs; and as proxies learn from past interactions, the context should be able to accumulate continuously.

Redis indicates that since its first release on Redis Iris in May 2026, customers have put it into production use.

Paradigm Chief Technology Officer June Lee said: "Everything you see in Paradigm is stored in Redis. It is the memory and orchestration layer that enables our agents to run in real-time and achieve a scale that was previously unattainable."

Redis Iris is now live. Intent Router, Agent Playbooks, Agent Memory, and Context Gateway are officially available; Lens and Agent Mesh are currently in the testing phase. For more information, please visit redis.io / iris.

Redis will hold its first online introduction event with its engineering team on November 12 (in the Americas) and November 17 (in Europe, the Middle East, and Africa). Registration can be obtained at redis.io / resources / videos / redis-iris-full-spectrum.

Redis indicates that Redis is the fastest data platform in the world. Since its open-source inception in 2011, and having become the most frequently cited first brand in caching solutions, Redis has helped over 10,000 customers build, scale, and deploy applications that power the modern world. By providing cloud and local databases for caching, generative AI, and other use cases, Redis helps digital enterprises set new standards for application speed.

Redis has offices in San Francisco, Austin, London, Tel Aviv, and Singapore. The company states that it has gained international recognition for its ability to rapidly build high-performance applications. For more information, please visit redis.io.

Message source: Redis