Poolside AI Acquires Fern Labs to Accelerate Long-Running Agent Capabilities
Fern Labs didn’t begin with the typical AI-startup playbook. No flashy demo, no rapid-fire feature drops, no race to ship an “agent” that could run for a few minutes before collapsing. Instead, Fern emerged with a quiet conviction: the next era of software wouldn’t be written by humans alone — and getting there would require agents that could think, revise, and persist over time.
Founded by Ash EdwardsDealroom has a profile for this one. Try Dealroom →, a former Palantir engineer steeped in systems thinking, Alex GoddijnDealroom has a profile for this one. Try Dealroom → and Taylor YoungDealroom has a profile for this one. Try Dealroom →, Fern set out to build something deceptively simple but notoriously difficult: an agentic operating system built for duration . Their architecture emphasized long loops of generation, critique, repair, and memory — the kind of structured, resilient reasoning cycles that most agent frameworks struggled to survive. In early prototypes, Fern’s agents didn’t just answer prompts; they completed multi-stage software builds, coordinated with specialized peers, and improved output quality the longer they ran. For enterprise teams watching the early agent hype fizzle, Fern’s pragmatism felt like a breath of fresh air.
That clarity of purpose caught the attention of Air Street CapitalDealroom has a profile for this one. Try Dealroom →, which quietly backed the company in a $3 million pre-launch round. With a small team and a tightly defined mission, Fern moved quickly, carving out a reputation as one of the few companies tackling hard problems — traceability, fault tolerance, multi-agent orchestration — rather than chasing hype cycles.
Which is why Poolside AI’s acquisition of Fern feels less like a surprise and more like the natural next chapter. Poolside has been building a frontier-scale stack for automated software development: large models trained for code, access to significant compute, and a push toward real enterprise deployment. What they didn’t have was Fern’s deeply opinionated agentic core — the scaffolding required to let such models operate continuously, reliably, and with context over long horizons.
Bringing Fern inside gives Poolside a crucial piece of its puzzle. And for Fern, it’s the platform they always needed: industrial-scale models, infrastructure, and distribution channels capable of taking their architecture from promising prototypes to production-grade systems.
Terms of the deal weren’t disclosed. But strategically, the logic is unmistakable. In a market crowded with short-lived agent demos, Poolside and Fern are now betting on something different: dependable, long-running autonomous agents that can write, maintain, and evolve software on their own — not in bursts, but over hours, days, and eventually, entire lifecycles. A big swing, but maybe the only one that matters.
Source:
Air Street Press
A.M.
