poolside transforms enterprises into AI-native, agentic organizations. We know first hand that building value-driving AI inside of large organizations takes more than a capable foundation model: it requires systems of agents that can operate in production environments and a high-touch motion with teams who know how to deploy them where failure has real consequences.
That’s why we’re excited to announce that poolside acquired Fern Labs, a London-based forward‑deployed research engineering company behind Bridge, a multi‑agent orchestration layer built for high‑stakes production environments.
With Fern Labs, poolside combines its frontier models and Model Factory infrastructure with on‑site agent deployments, giving customers a full stack to design, prove, and scale real agentic systems entirely within their security boundary.
What Fern Labs built
Fern Labs was founded by Ash Edwards, Alex Goddijn, and Taylor Young after leading complex deployments at Palantir. Fern’s work centers on long‑running, multi‑agent systems that understand enterprise environments and can be debugged, audited, and improved quickly. With their forward-deployed research engineering approach, the team embeds directly with customers to co‑design the workflow, prove value on a high‑stakes use case, and translate lessons back into product.
The agentic capabilities of frontier models continue to rapidly improve, offering radical new approaches for many enterprises' most critical processes. As a result, mastering how to get agents to work at scale, how to make multi-agent systems stable for tasks that matter, and how to break problems down effectively is both art and science. Fern Labs built a culture around forward-deployed research engineers who can do both the engineering and the R&D on the front line.
How poolside and Fern Labs will accelerate together
poolside deploys frontier AI systems within enterprise boundaries and in regulated industries. With Fern Labs’ software and expertise, we’re able to accelerate the deployment of agents inside private customer environments, observe their problems in context, translate those into reinforcement learning-shaped problems, and train models to perform exceptionally well for those specific use cases, all using our existing infrastructure.
To truly deliver value for customers, you need three things: a clear understanding of the customer’s problem space; deep knowledge of agents and how to structure multi-agent systems; and the tools to experiment quickly and deploy securely.
Together with poolside’s industry leading GTM team, this acquisition accelerates our expansion of agents within the world’s highest-consequence environments.
How our customers will benefit
Pairing forward-deployed research engineers and Bridge with our models, poolside can now embed directly with clients to build and deploy custom high-performance agentic systems against their most valuable problems. Our engagement model is as follows:
- Land on one high‑stakes problem. Typically, this is a core operational workflow, like resolving incidents faster, detecting quality issues earlier in production, or automating customer onboarding, for example.
- Build an agentic system with Bridge using poolside models, measurably reducing time‑to‑resolution, reducing cost, increasing throughput, or increasing revenue. The system is embedded within the organization, with humans in the loop at key decision points.
- Expand to adjacent use cases; each deployment makes the next one faster, as new use cases build upon the previously configured environments, processes, and agent building blocks. Over time, you create a scalable agent workforce that creates compounding value.
Building toward the agentic enterprise
Having multi-agent systems operate reliably in high-stakes environments remains an open problem. There’s a lot to figure out: what are the right patterns for how agents cooperate? How do agents handle uncertainty, and how do they recover from failure when operating unsupervised? How do we allow agents to run for hours, or soon days, without sacrificing observability and robustness?
All of this will happen at the edge, where agents meet real constraints like data security, for example. Cracking this requires engineers who can both build the AI systems and understand the domain. It requires a product that supports rapid iteration without breaking what's already deployed. And it requires a model that improves based on what those agents learn. poolside now has all three: Bridge for orchestration, a team that's proven they can deploy in secure and highly-regulated environments, and a model that improves with every deployment.
Getting there requires deploying at scale and learning from those deployments, to bridge the gap between lab research and real enterprise value. We’re super excited to have Fern Labs join the team at a key part of poolside’s growth.