Fivetran outlines an open, autonomous future for AI agents

Agents need access to tools and business data

In a November 11, 2026 letter, Fivetran CEO and Co-founder George Fraser, President and Co-Founder Tristan Handy, and Co-Founder and COO Taylor Brown described the current shift in computing as an “age of intelligence.” They compared it with earlier changes driven by the PC, internet, mobile, social and cloud technologies, while arguing that this shift will be larger because it could change both the role of firms and the role of people in production.

The founders define an agent as intelligence combined with tools and data. Agents can have goals, ask and answer questions, and perform actions, they wrote. They say agents are already effective in knowledge work where context is available and outcomes can be verified, including software engineering.

The letter contrasts a contract review, which may require a contract, redline history, email exchanges and contracting guidelines, with a business investigation that could require structured data, departmental OKRs, meeting notes and information about current initiatives. According to Fivetran, the second type of work has a far larger data requirement, even when its results can be evaluated.

From more analysis to autonomous processes

Fivetran says time limits leave human analysts unable to explore every possible explanation behind complex business questions. It also argues that many questions are never asked because organizations lack enough analyst capacity. The founders say organizations that deploy agents have unlocked “50x more question / answer loops,” because agents can work continuously and be available immediately.

The company expects existing infrastructure, including compute, networking, programming languages, APIs and data infrastructure, to provide the foundation for agents. It also identifies areas that still need development, including evaluations and background memory formation. Fivetran calls this emerging “agent stack” a way for agents to maintain a shared picture of a business and add new discoveries to a continual learning loop.

Fivetran's near-term vision is AI helping knowledge workers accomplish more. The longer-term direction is autonomous agents that own business processes end to end. In its customer support example, an agent would resolve a ticket, monitor the customer's account, reach out when needed and file issues directly with a software engineering agent for broader fixes.

Fivetran calls for open infrastructure and human oversight

The founders describe “agent engineers” as people responsible for redesigning processes around agents, connecting them to tools and data, evaluating their performance, and managing organizational change. These responsibilities span process engineering, systems engineering, evaluation and operations, and change management. Fivetran says companies that create room for these roles will see the greatest impact from AI.

The letter argues that companies should own their data and governance and be able to change models and compute providers. Fivetran says this makes open source, open standards and cross-platform portability strategically important, and links the position to its commitment to building its platform around Open Data Infrastructure principles.

Fivetran says analytics agents are already here and autonomous agents are coming. While it plans to build the infrastructure for them, the company also says human qualities such as empathy, goal-setting and innovation will become more valuable as AI capabilities grow. Its stated aim is to continue investing in the people who build and use this agentic future.