Couchbase, has announced the general availability of the Couchbase AI Data Plane, a unified data infrastructure layer for enterprise artificial intelligence (AI) agents.

The platform is designed to give enterprises persistent agent memory, real-time context retrieval, and consistent data access across cloud, edge, and lakehouse architectures. By bringing together data services that are often deployed as separate point solutions, Couchbase says the AI Data Plane can help enterprises move agentic AI from pilots to production with more consistent decision-making, richer customer experiences, and measurable efficiency gains.

The AI Data Plane unifies Agent Memory, an Agent Catalog for discoverable agent tooling, and an enterprise-supported, self-managed Model Context Protocol (MCP) server for standardised integration. It consolidates earlier Couchbase deployment models into a single architecture that runs across Couchbase Capella and self-managed environments.

The launch is also supported by Enterprise Analytics 2.2, which adds Apache Iceberg-based lakehouse federation, along with a Trino adapter expected in Q3 2026. Couchbase says the platform is backed by its engineering and support organisation, giving platform teams a single operational surface for the data services on which AI agents depend.

“Most enterprises quickly discover that moving from chat-style pilots to production-grade agentic systems is really a data problem, not just a model problem,” said Devin Pratt, Research Director, AI, Automation, Data & Analytics, IDC.

According to IDC, 80% of agentic AI use cases will require real-time, contextual, and widely accessible data. Pratt said architectures that make agent memory and context retrieval first-class database capabilities can reduce the integration burden that has slowed real-world agent deployments.

Persistent agent memory

For chief information officers (CIOs) shaping their AI infrastructure strategies, the challenge is no longer only about choosing models. It is increasingly about governing memory, context, and retrieval across the full agent lifecycle.

Couchbase Agent Memory addresses this by providing a unified persistence layer inside the operational data platform. Instead of forcing teams to stitch together separate caching, vector, and document stores, it gives developers a single service for agent memory and retrieval.

The layer is framework-agnostic and has been validated with LangGraph, CrewAI, and LlamaIndex. This allows engineering teams to switch or combine orchestration frameworks without rebuilding the memory layer.

As enterprises move from prototypes to production agents, the gap between what agents can reason about and what they can remember across sessions has become a major bottleneck. Simple agents may work with vector search alone, but production-grade agents need to store conversational context, retrieve structured operational data, and maintain state across sessions and restarts.

Couchbase says the AI Data Plane supports these requirements with sub-millisecond latency at the point of decision.

“What matters most for enterprise-grade conversational AI agents is that data retrieval is fast, consistent, and seamless. When you are running human-to-AI agent interactions, everything behind the scenes needs to be predictable and consistent to provide natural interaction,” said Patrick Ferriter, SVP of Product, Agora.

“That is what we are solving together with Couchbase, and it is why we chose them as a partner for the data layer for our conversational AI platform. Every one of our conversational AI use cases requires efficient data retrieval to feed the pipeline for AI agents, whether that is outbound sales, customer service, physical AI, or something entirely new. We have had a multi-year relationship with Couchbase, and as we have scaled into agentic workloads, this was a natural extension to our partnership,” he added.

Built for agentic AI at the edge

The shift from single-prompt AI applications to multi-step autonomous agents has exposed a mismatch between how agents work and how most data infrastructure is built, especially at the edge.

Agents operate across sessions, build context over time, and need to act on structured operational data and unstructured embeddings at the same time. They also need this capability across cloud, edge, and devices.

The Couchbase AI Data Plane has been architected to support this distributed data requirement. Every agent action may trigger context retrieval, memory writes, and state synchronisation, often across thousands of concurrent sessions. Couchbase says the platform uses its scale-out, memory-first architecture to support high-throughput workloads with low latency.

The AI Data Plane builds on Couchbase’s distributed multi-model architecture, which supports JSON documents, key-value data, SQL for JSON queries, full-text search, eventing, and vector search in a single distributed system.

Agent Memory extends this foundation with session persistence and context retrieval, while the MCP server and Agent Catalog provide the integration and observability layers needed for production agent deployments.

Enterprise Analytics 2.2 and lakehouse federation

Couchbase also announced Enterprise Analytics 2.2, an expansion of its analytics capabilities. The release opens operational data in Couchbase to the broader lakehouse ecosystem while strengthening the query engine.

Enterprise Analytics 2.2 introduces Apache Iceberg lakehouse federation. This allows teams to query real-time operational analytics from Couchbase alongside existing Iceberg-based lakehouse tables without complex extract, transform, and load (ETL) processes or data duplication.

The capability is aimed at enterprises adopting Iceberg for open governance, performance, and ecosystem flexibility. By connecting Iceberg tables with the same platform serving agentic workloads, Couchbase says organisations can derive more value from their lakehouse investments.

The analytics update also adds Google Cloud Storage support, JSON Web Token (JWT) authentication, Oracle and SQL Server change data capture, asynchronous long-running queries, an index advisor, index-only query plans, SQL++ UPDATE support, and software development kit (SDK) updates across Java, .NET, Python, JavaScript, and Go.

A new Trino adapter, expected in Q3 2026, will provide in-place SQL access to Couchbase operational data from Trino-based platforms, including AWS Athena, Amazon EMR, Google Dataproc, and Starburst. This is intended to reduce the need to extract and replicate live data into separate analytical stores before querying it for AI and analytics workflows.

Capella iQ gets policy-led model selection

Capella iQ, Couchbase’s natural-language query assistant, now supports multi-model provider selection with AWS Bedrock and OpenAI.

The capability is governed through organisation-level policies. Administrators can control which models are available to which teams, helping keep inference costs and data residency requirements within enterprise guardrails.

This gives developers the flexibility to choose the right model for each workload, while allowing administrators to maintain central control over compliance, cost, and data residency.

Edge, mobile, and distributed application updates

As AI agents become part of operational workflows, their data requirements are moving closer to the point of work. This increasingly means devices, field environments, and network-edge locations rather than only centralised data centres.

Couchbase is extending the AI Data Plane to the edge so that agents running in mobile and edge environments can access replicated data and perform local vector search, even when connectivity is intermittent or unavailable.

The production bet

For Couchbase, the launch is positioned around a larger enterprise shift. As companies move beyond AI pilots, the database layer becomes central to whether agents can scale safely and reliably.

“The database layer is where agentic AI either scales or stalls, and most of the industry is still treating agent memory as an afterthought,” said Barry Morris, Chief Product and Strategy Officer, Couchbase.

“We built the AI Data Plane because our customers told us that stitching together separate vector, caching, and document stores for every agent was the single biggest drag on their production timelines. Agent Memory gives them a unified, framework-agnostic persistence layer that operates identically in cloud and self-managed environments from cloud to edge, and runs at the latency their agents actually need. That is what it takes to move from pilot to production, and the vendors who understand this will define the infrastructure category for the next decade of AI,” he added.

For enterprises, the larger message is clear. Agentic AI will not move into production on model capability alone. It will need a governed data layer that can provide memory, context, retrieval, synchronisation, and performance wherever the agent is expected to act.