Fundraise

Swamp Club raises $7M seed to fix AI agents' cost and reliability problems

What's the deal? Swamp ClubDealroom has a profile for this one. Try Dealroom → has raised $7 million in seed funding to launch SwampDealroom has a profile for this one. Try Dealroom →, a platform for building AI agents through what it calls behavior-centric design. Amplify Partners led the round, with Encoded VenturesDealroom has a profile for this one. Try Dealroom → participating.

What's the endgame? Swamp aims to replace the prevailing "agent-centric" approach, where large language models decide an application's control flow at every step. Instead, it uses frontier models to translate a developer's intent into reusable, deterministic behaviors that "run the same way every time."

Why now? The company argues that current agent designs are too fragile and expensive to scale in production. Gartner predicts over 40% of agentic AI projects will be cancelled by 2027, while McKinseyDealroom has a profile for this one. Try Dealroom → found only 37% of enterprises using AI can show positive EBIT impact.

By the numbers: Swamp says agents built on its platform use less than one-tenth the tokens of rival frameworks and run up to six times faster. In a published comparison, a trip-planning agent built with LangChain consumed 24,611 input tokens; the same agent on Swamp used 1,183 — roughly 14 times fewer, at one-sixth the cost.

Who's using it? Accounting software firm Xero is an early customer. "Work that used to take hours or days now takes minutes," said Adam Roberts, its senior vice president of data science, analytics, and insights. "We started with one prototype for marketing, and now other teams at Xero are building on the same platform."

The signal: At $7 million, Swamp's seed lands in the top 4% of US media-sector seed rounds on record, a sign of investor appetite for infrastructure that makes AI agents cheaper and more reliable. As the industry shifts from experimentation to production, the fight is moving from model capability to the architecture that runs it.

Read more: morningstar.com

Image credit: Generated with Gemini

More top stories