Couchbase, which has undergone a nearly complete leadership change since being acquired and taken private in 2025, says its new AI Data Plane provides a single, governed data layer—including persistent agent memory, real-time context retrieval—for AI agents running in production.
In a move to unify the fragmented data services that often derail agentic AI projects, Couchbase Tuesday announced the general availability of its AI Data Plane, new software that the next-generation database developer said provides a single, governed data layer for AI agents running in production.
The new Couchbase AI Data Plane provides enterprise AI agent systems with persistent agent memory, real-time context retrieval, and consistent data access across cloud, edge and lakehouse systems, according to the company.
“I think everybody shares the opinion that at this point in this world of AI, data really needs to be a first-class citizen in any AI system,” said Tim Finley, Couchbase senior vice president of AI, in an interview with CRN.
[Related: Couchbase Launches Database AI Services That Unify Data With AI Models]
The AI Data Plane announcement comes at a pivotal time for Couchbase, headquartered in San Jose, Calif., which was acquired last year by private equity firm Haveli Investments in an all-cash transaction valued at $1.5 billion. (The deal, initially announced in June 2025, was completed on September 24.)
Since then, almost the entire Couchbase executive team has changed. Matt Cain, president and CEO of Couchbase for more than eight years, stepped down shortly after the acquisition and was replaced by BJ Schaknowski, previously chief executive at healthcare operations platform developer Symplr, as CEO.
At the same time Amir Jafari was appointed the new CFO. That was followed in December with the appointments of Deirdre Toner as president and commercial officer and Barry Morris as chief product and strategy officer. Gopi Duddi, hired in 2022 as senior vice president of engineering, was promoted to chief technology officer in December.
The new Couchbase AI Data Plane is the latest example of Couchbase’s overhauled product strategy that focuses on mission-critical applications, infrastructure consolidation and providing AI-ready data.
Couchbase’s AI Data Plane news also comes at a pivotal time for the “big data” sector of the IT industry. AI agents and applications require huge amounts of data to function and in recent months many database and data management tool vendors have been adding capabilities to their products to meet those needs.
But Finley said many organizations are supporting their AI initiatives with legacy data infrastructure, often made up of traditional transactional databases, NoSQL databases, vector stores, search engines and separate caches—all lashed together.
“AI, in general, needs more than a model,” Finley said. “Frameworks and models don’t quite get the job done. Business enterprises, in particular, need more than a model.”
“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, IDC research director of AI, Automation, Data & Analytics, in the Couchbase press announcement.
AI Agent Data Requirements
IDC expects that 80 percent of agentic AI use cases will require “real-time contextual, and widely accessible data,” making agent memory and context retrieval key capabilities for the databases that support agentic AI.
“It all comes down to data, and data is either the problem or it is the solution,” Finley said. Missing or unreliable data context often hinders AI projects, he said, as does data scattered across multiple silos.
While many businesses and organizations recognize their data dilemma, Finley said, “they haven’t done anything about it yet, and that is because legacy infrastructure can’t do it.”
Couchbase has been positioning its core database as a key component of enterprise-scale AI initiatives with its flexible JSON architecture and such capabilities as high-performance indexing, vector search, real-time analytics, graph data management, and more.
Couchbase’s position is that IT managers wrestling with AI infrastructure strategies need a unified data platform that governs memory, context and retrieval across the full agent lifecycle.
Organizations often put a great deal of focus on large language models (LLMs) as they develop and implement AI systems, often overlooking the data components, Finley said.
“We love models, but models are rentable, literally,” he said. “That’s how people get access to models. What’s important is what people own, and that is their data, and their data is strategic. And so we give them a way to get value out of that data for the enterprise, securely, safely, reliably, and predictably. In the simplest sense, it’s the [AI Data] Plane that puts it to work.”
AI Data Plane Details
The AI Data Plane unifies operational data, vectors, documents and cache in a single platform and consolidates previous Couchbase deployment models into a single architecture that runs on the company’s popular Couchbase Capella fully managed database-as-a-server and on the Couchbase Server database in self-managed environments. (Some of the capabilities in AI Data Plane were previously available as “Capella AI Services.”)
AI Data Plane provides persistent agent memory, an agent catalog for discoverable agent tooling, and a self-managed MCP server for standardized integration of model context protocols.
AI Data Plane lowers business process latency and AI token expenses by reducing “round trips” between agents and LLMs, said Finley, who called the new offering “category defining.”
Finley said AI Data Plane creates opportunities for solution providers and system integrators by offering a solution to replace kludgy AI systems with a single platform to which partners can add their own intellectual property and industry expertise.
“Because, ultimately, agentic flows are just business processes,” Finley said.
In May Couchbase and integration platform developer Boomi struck an alliance through which the two companies are co-engineering systems that provide a production-ready foundation for agentic AI.
“Agentic memory—both historical and dynamic, both individual and shared—is one of the least mature capabilities in the AI landscape, but also one of the highest potential areas,” said Matt McLarty, Boomi CTO, in an email exchange with CRN. “Couchbase’s AI Data Plane offers unique functionality and scalability that makes it a tremendous option for agentic memory, especially in combination with Boomi’s data activation and agentic orchestration capabilities.”
Couchbase also unveiled what it called a major expansion of its Enterprise Analytics 2.2 capabilities that connect operational data in Couchbase to lakehouse ecosystems with support for Apache Iceberg-based lakehouse federation. Couchbase also plans to provide an adaptor for the Trino SQL query engine in the third quarter. And the company has introduced enhancements to Capella iQ, a natural language query assistant, that now supports multi-model provider selection with AWS Bedrock and OpenAI.
