AI models and infrastructure are now good enough for the enterprise, yet few teams are seeing repeatable, production-level success with agents. This is because agents lack the right context, at the right time, remembered the right way. Those that do deliver value often do so inefficiently, with over-stuffed prompts, redundant LLM calls, and brittle custom integrations that cost more in tokens and time than the value they return.
Here is how I think about it: agents are software. They need memory, context, routing, communication, and access to live application data, and getting all of that shouldn't require gluing multiple different infrastructure products together. Redis already sits in the middle of your application. Redis Iris adds those agent primitives right alongside your live data, so your agents work reliably in production and at scale.
Today, I'm excited to share the next chapter of Iris, our real-time context engine for AI agents with you.
Six capabilities, one runtime
Redis Iris now offers six customizable capabilities. Use them together, independently, or in any combination, and tune each to build production-ready agents from day one.
The Redis Iris Reference Architecture: one engine, six capabilities
- Intent Router (GA) matches the meaning of a request rather than its wording, then sends it to the right destination: a remembered FAQ, an agent skill, a human, or an LLM. It also handles agent and model routing, guardrails, and “fuzzy” if-statements, so decisions aren't limited to true or false.
- Agent Playbooks (GA) lets you define which answers are stable and should be reused, and which need current data and a fresh run of a known procedure. Playbooks get more thorough and accurate the more they run, which cuts resolution times and token spend.
- Agent Memory (GA) builds memories from user actions, documents, and other semi-structured data through MemoryETL. It organizes them in hierarchies for finer-grained control and gives you inspection, audit, and provenance. It can also forget what it shouldn't keep, such as personally identifiable information.
- Context Gateway (GA) gives agents a single, governed point of entry to connected context sources. Bring your own MCP servers, cache which tools solve a given problem to reduce token use, and build reusable results for consistency across sessions.
- Lens (beta) makes agent behavior traceable. Every outcome can be traced back to the skill, route, or FAQ that served it, the memory it learned from, and the source session. You get auditable evidence, plus visibility into task success, tokens per task, and latency.
- Agent Mesh (beta) lets agents communicate over ephemeral streams, with automatic buffering and reconnection when a connection drops. Durable SSE Streams keep responses flowing through interruptions, and A2A Comms give agents a reliable channel for delegation and handoffs.
Together, the six capabilities help ensure agents remember the right things, answer the right way, and are transparent in every action, all while delivering more outcomes per token. And because Iris runs on the Redis technology your team already knows and loves, you can cut the custom integration work that usually stitches together retrieval, memory, routing, and messaging.
Agents that survive the real world
Production agents have to handle unpredictable human inputs and behavior, multiple channels, and handoffs, any of which can become a point of failure. Iris creates the layer of the AI stack that sits between an agent and the data and tools it needs to act, and is built around four principles:
- Navigable: Context delivers navigable systems, not silos. The best outcomes happen when agents can follow connections between the available data and tools.
- Adaptive: Good agents know what to remember and what to forget. Remembering gives agents continuity, and forgetting keeps context relevant and governable.
- Efficient: Efficient context keeps costs in check. Agents that retrieve only what’s relevant and reuse the answers and procedures they’ve already generated cost less to run.
- Compounding: The context agents use must learn continuously and grow over time.
Put together, these principles are what Iris delivers: agents that find what they need, keep what matters, cost less to run, and get better with every interaction.
Iris first launched in May 2026, and customers have been putting it to work in production ever since, including June Lee, CTO of Paradigm: “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.”
Build production-ready agents today
The models are ready, and the next chapter of Redis Iris is here. With Iris, the conversation shifts from “Can we build an agent?” to “Let's operate an agent that is reliable, explainable, and increasingly valuable over time.” Now your agents have the ability to remember the right things, answer the right way, and be transparent in every action, all while delivering more outcomes per token.
Ready to get started? Pick one business-critical use case and build it. Get started with 5,000 free messages.
Ready to learn more? Read ourtechnical deep dive from Simba Khadder, Head of AI products, watch his demo video, and register for our next webinar.
See Iris in action
Join us for a live look and demos on Nov. 12 (Americas) or Nov. 17 (EMEA).
