Dune cuts 25% of staff to focus on AI agents for finance

On May 14, Dune Analytics announced it had reduced its workforce by 25% as part of a company-wide restructuring led by cofounder Fredrik Haga. The change refocuses the platform on AI agents and services for institutional on-chain finance, including tokenized stocks, bonds and commodities.

Haga wrote that the company will keep its end-to-end data stack while narrowing its product focus. He characterized the departing employees as “exceptional people” and offered recommendations to other employers. The company has raised about $79 million in total funding, including a $69.4 million Series B in 2022.

The restructuring centers on Dune MCP, an open-standard server Dune launched in March 2026. MCP lets AI agents query the platform’s data warehouse using natural language and includes a set of about a dozen tools for table discovery, query execution and visualization across more than 100 blockchains. Dune also released a dbt Connector intended for teams building on-chain data pipelines to support automated, programmatic workflows instead of manual dashboarding.

Haga described the change as a tightening of product focus. In his post he wrote, “We’re restructuring Dune to sharpen our focus around the core data products thousands of customers across the crypto industry rely on.”

The announcement drew a public response from Ryan Li, cofounder of Surf. Li, who said he was a former heavy user of the platform, replied that crypto research now needs infrastructure built for AI agents rather than human-operated dashboards. He wrote, “operating in the AI era demands infrastructure built for agents, not humans clicking through dashboards.” Surf raised $15 million in December 2025 from investors including Pantera Capital, Coinbase Ventures and Digital Currency Group.

The exchange between the two founders came amid competition in crypto data infrastructure as established analytics platforms adjust to AI-driven workflows and newer entrants promote agent-native stacks. Dune’s MCP and dbt Connector are designed to make on-chain data more accessible to automated systems, while other firms emphasize fast query engines, reliable SQL and structured outputs for programmatic use.

The company wrote that it will continue to support existing customers and core products while offering dedicated services to institutional clients that are tokenizing traditional assets or integrating on-chain data into conventional finance operations.