Couchbase adds agentic AI development suite to Capella DBaaS

Fueling development

Agents first emerged as the cutting edge of AI development in mid-2024.

For existing Couchbase customers, AI Services are important because they turn Capella from a database that supports AI into a place where operational and analytical data, models and agents are designed to live together.

Enterprises began substantially increasing their investments in AI projects following OpenAI's November 2022 launch of ChatGPT, which marked a significant improvement in generative AI (GenAI) technology. Through mid-2024, those investments largely focused on building GenAI chatbots that enable users to query and analyze data using natural language as well as automate certain tasks when prompted, such as generating code and documenting work.

Once agentic AI emerged, investments switched to building agents rather than chatbots. Agentic AI tools, unlike GenAI applications, are capable of autonomous action, such as analyzing data to surface insights and performing business processes without being prompted.

In response to surging interest in AI development, many data management vendors created environments that simplify the complex process of building AI tools trained on proprietary data so that applications understand the unique characteristics of individual enterprises. Just as enterprises initially focused on GenAI following ChatGPT's launch, data management vendors' initial AI development suites were similarly focused on GenAI. Over the past year, as agentic AI has replaced GenAI, vendors have added capabilities that simplify building agents.

Databricks, Snowflake, Teradata, Alation and Informatica are some of the many data management vendors providing agentic AI development environments. So are the tech giants, with Oracle the most recent addition.

Now, a year after initially introducing Capella AI Services, Couchbase AI Services is GA for the vendor's customers. Customer feedback is helping provide Couchbase the impetus for building its own agentic AI development suite, according to Rahul Pradhan, the vendor's vice president of product and strategy.

"Our customers were already using Couchbase as the operational system of record for their most critical applications, and they were experimenting with generative AI and agents on top of that data," he said. "What they kept telling us was that the AI stack felt fragmented and fragile … and [they needed] a lot of custom glue code to make it all work in a secure, governed way."

In addition, observing market trends motivated Couchbase to develop AI Services, Pradhan continued.

"We saw that the next wave of AI won't just be about chatbots -- it will be about agents that need long-lived memory, context and real-time access to operational data," he said. "AI Services is our way of bringing those pieces together."

Collectively, the features included in AI Services comprise a complete set for building agents, according to Pratt.

Key elements of an agentic AI development framework include secure model hosting, data processing of structured and unstructured data, vector database capabilities, and governance.

"Couchbase brings the critical ingredients for enterprise agents into one platform, giving users an end-to-end environment to develop, govern and scale agentic applications on Capella," Pratt said.

Perhaps the most significant component of AI Services is the governance layer provided by Agent Catalog, he continued.

Aslett, meanwhile, highlighted the integration with Nvidia, including support for the Nvidia NIM suite of microservices.

"Couchbase's support for Nvidia NIM, for example, is important in enabling users to accelerate the development and high-performance operationalization of AI models," he said.

One key agentic AI development feature not mentioned by Couchbase as part of its AI Services launch is support for Model Context Protocol, a framework for securely connecting agents with external data sources such as large language models that was developed by AI provider Anthropic in November 2024.

However, though not mentioned by name, Couchbase does provide an MCP server and plans to make MCP easier to use in conjunction with other AI development tools on the Capella platform as it improves AI Services, according to Pradhan.

"For this GA announcement, we chose to keep the message tight around the core value of Couchbase AI Services rather than spotlighting any one protocol," he said. "MCP is [important], but it's one piece of a broader story. Looking ahead, customers are clearly interested in more 'managed' MCP experiences … and we're actively evaluating options to make MCP even easier to use with Couchbase."

Looking ahead

With AI Services now part of the Capella platform, Couchbase has wide-ranging plans for 2026.

Initiatives include developing more advanced retrieval techniques to help users discover relevant data within complex datasets, expanding to become a memory layer for agents so they can reason and gain historical context, simplifying using Capella to improve developer productivity and improving its governance capabilities, according to Pradhan.

"Our overarching goal is simple -- make it dramatically easier and safer for enterprises to build real, production agentic applications on top of Couchbase," he said.

Aslett, meanwhile, suggested that Couchbase make messaging one of its focal points. Particularly, showing how current users are successfully developing agents on Capella could help Couchbase appeal to potential new ones.

"Potential customers are likely to be looking for … practical examples of how AI Services is being used to accelerate the development of agents and how its capabilities compare to those provided by specialist agentic and generative AI development tools and platform providers," he said.

Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than 25 years of experience. He covers analytics and data management.