Dune Cuts 25% to Pivot Toward AI Agents

Dune Analytics announced on May 14, 2026, that it has cut roughly 25% of its staff as part of a strategic restructuring to concentrate on AI agents and institutional on‑chain data products.

Co‑founder and CEO Fredrik Haga posted the change on X, writing that Dune will “sharpen our focus around the core data products thousands of customers across the crypto industry rely on,” and that the company had let 25% of the team go this week.

The reorientation leans heavily on agent‑native tooling Dune has rolled out this year, most notably Dune MCP — an open‑standard Model Context Protocol server that gives AI agents structured, programmatic access to Dune’s data warehouse across more than 100 chains.

Dune has also released a dbt Connector, a CLI and a Skills framework since March, moves the company says will let teams and autonomous agents discover tables, run SQL queries, and render charts without hand‑built ETL stacks. Those product shifts are central to the company’s stated push to be “agent‑friendly.”

Haga told followers Dune remains well capitalized and will preserve its end‑to‑end data stack while offering expanded white‑glove services for institutional tokenization projects, a line that signals a tilt toward customers who need contract, regulatory, and audit controls.

The announcement drew immediate commentary from other builders in the space. Surf co‑founder Ryan Li replied that crypto research “demands infrastructure built for agents, not humans clicking through dashboards,” framing the move as part of a competitive race between traditional dashboard vendors and agent‑native entrants.

Dune’s cuts reflect a broader pattern in crypto and tech: several firms this spring paired layoffs with declarations that AI is changing how work gets done. Coinbase, for example, cut about 14% of its workforce in early May as it said it would reorganize into smaller, AI‑native teams.

Industry observers say the pattern mixes cost‑management with a bet that software agents will automate many research and analyst workflows, reducing headcount needs while increasing demand for high‑quality, machine‑readable data. That dynamic helps explain why analytics platforms are retooling interfaces, output formats, and security controls to suit automated consumers as well as human analysts.

For affected employees the change is immediate. Public reporting and company posts suggest Dune employs roughly 150 people, so a 25% reduction likely impacts several dozen roles; Haga said he’s recommending strong performers to other firms and flagged severance and support in his post.

For customers and integrators, the shift could speed adoption of agent‑driven workflows: Dune’s MCP returns structured results agents can consume programmatically, while connectors and skills aim to shorten the path from prompt to production analytics. That can reduce time‑to‑insight for institutional trading desks or tokenized‑asset teams.

But the pivot raises questions about product scope and market timing. Some customers still rely on dashboards and bespoke analyst workflows; making those tools secondary could frustrate parts of Dune’s longtime user base even as new institutional deals are pursued. Financial and regulatory demands from institutional clients will also force heavier investment in access controls, SLAs, and compliance features.

The move is a concrete example of how crypto analytics providers are repositioning for an “agent era” where autonomous systems — not just people — will query blockchains at scale. For Dune, success will hinge on turning agent‑friendly primitives into reliable, institution‑grade services that justify the personnel tradeoffs.