Redis launched its Iris platform Monday, targeting enterprise AI agents that require faster data retrieval than traditional retrieval-augmented generation (RAG) systems can provide. The platform combines real-time data ingestion, semantic interfaces, and agent memory capabilities built on Redis Flex, a storage engine that runs 99% of data on flash storage at reduced costs compared to in-memory solutions.

The announcement reflects broader enterprise challenges with AI agent infrastructure. According to VentureBeat's Q1 2026 market data, buyer intent for hybrid retrieval systems increased from 10.3% to 33.3% between January and March, while custom in-house retrieval implementations rose from 24.1% to 35.6% as companies outgrew standard options.

Redis CEO Rowan Trollope said the core issue is scale mismatch, noting that agents generate orders of magnitude more data requests than human users. The Iris platform addresses this through five components: Redis Data Integration for continuous data syncing, Context Retriever for semantic data access, Agent Memory for session state management, Redis Flex storage engine, and Redis Search with LangCache for semantic caching.

Separately, Anthropic introduced new security features for Claude Managed Agents, including self-hosted sandboxes and MCP tunnels. These capabilities allow enterprises to run tool execution within their own infrastructure while keeping agent orchestration on Anthropic's platform, addressing credential security concerns that have limited enterprise AI agent adoption.

The self-hosted sandbox feature, currently in public beta, separates the agent's decision-making process from tool execution, while MCP tunnels in research preview connect agents to private servers without exposing authentication credentials. OpenAI previously added similar local execution capabilities to its Agents SDK.

Stephanie Walter of HyperFRAME Research noted that the market is converging on the need for "governed, current, low-latency context" for AI agents. She emphasized that successful context layer technologies must make agents "faster, cheaper, and safer to run" while addressing enterprise governance requirements for data access and cost management.