Couchbase Makes Case for Agent Data Foundation

Couchbase Makes Case for Agent Data Foundation
Wednesday, November 18, 2026
  • •Couchbase’s Tim Finley said company data supplies business context that can make AI results more relevant and reduce unnecessary model calls.
  • •Finley cited agents’ more than 60% share of web traffic and 450% more network traffic than humans doing the same tasks.
  • •Couchbase reported 40% to 60% fewer LLM calls in semantic caching tests and is evaluating roughly 110 tools and skills.

In an interview published November 17, 2026, Couchbase’s Tim Finley told Constellation Research’s Holger Mueller that company data can give AI systems useful business context and distinguish one company’s results from another’s. He argued that organizations using the same AI models can still produce more relevant results by grounding them in their own customer, operational, and business knowledge. Better context may also reduce costs by avoiding unnecessary model calls.

Finley said that cost pressure increases as AI use shifts from occasional prompts to always-on agents. He cited a Cloudflare analysis reporting that autonomous agents and machines account for more than 60% of web traffic, 18 months ahead of schedule, and a Cisco study finding that agents generated 450% more network traffic than humans doing the same tasks. Those figures were cited during the interview as evidence of infrastructure demands organizations should consider as agent workloads scale.

Finley described Couchbase’s approach as a converged data platform that combines a scalable JSON database with SQL capabilities. He said the company’s AI investments include vector search, model services, LLM functions, semantic caching, and agent memory. In an integration with Amazon Bedrock, semantic caching reduced LLM calls by approximately 40% to 60% in Couchbase’s testing; fewer calls can help lower costs and latency while preserving access to capable models.

Couchbase launched its AI Data Plane in June, Finley said. It brings together model services, agent memory, and an agent catalog. He also cited an early customer that put multiple agents into production within weeks of general availability. Customers have shown strong interest in Couchbase’s Model Context Protocol (MCP) offering, he said, and the company is evaluating roughly 110 tools and skills developed by field engineers for possible future releases. Finley suggested that tools making data capabilities easier for agents to use can also help the people who work with them.

As agent activity grows, Finley said, organizations need to monitor tool calls, handoffs between agents, model access, and token usage. Couchbase can capture this activity, which he called “digital exhaust,” in JSON that teams can query with SQL to investigate what is succeeding or failing. The company’s longer-term goal is full provenance at scale, so organizations can determine what an agent knew, what it did, and what resulted from each action. Finley said Couchbase aims to deliver this within the next calendar year. He argued that governance need not delay agent development: organizations can build with governance in mind and select infrastructure designed to support visibility and control as workloads grow.

Subscribe to the AIB Newsletter

The fastest, most reliable AI news, straight to your inbox

Companies related to this article

Blog posts related to this article