Our investment in Fern Labs

$3M day one financing to build multi-agent systems for real-world problems.

Fern’s founding team: Alex Goddjin, Ash Edwards, Taylor Young.

Agents for open-endedness

At Air Street, we’ve long been enthusiastic about the potential of open-ended systems to tackle economically valuable tasks. The stunning results OpenAI displayed with o1 and o3, and now DeepSeek’s R1, combined with the plummeting cost of inference suggests that AI will be used to tackle increasingly complex and ambitious tasks.

While agentic AI applications can display stellar results in the lab, they usually have a patchier record when they’re deployed on real-world tasks. Especially if you’ve not up-skilled as a model whisperer.

As we have seen, scaling test-time compute can lead to markedly better performance for LLMs. The key to success will lie in providing agents with safe, flexible environments where they can collaborate at scale, without detaching them from the actual challenges they need to solve.

We’ve led a $3M day one financing for Fern Labs

I’m happy to share that Air Street has led a $3M day one financing in Fern Labs.

Fern Labs is building a platform to scale coordinated agent compute against hard problems. By deploying networks of agents, Fern today can create and test software, perform small data science and analysis projects, and generate and refine resources like documents and presentations. As Fern and the base models improve, the scope of solvable tasks increases, with the end goal of Fern performing large-scale projects and companies.

You can see Fern in action here, building its own version of Asana for an agentic world. Not only does Fern write the front and back ends, the site is deployable and usable by both humans and agents alike.

Branching out from Palantir to Fern

The founding team - Ash, Alex, Taylor - worked on some of Palantir’s first agent-driven AI systems. Thanks to the company’s forward deployment model, they also know what it takes to navigate the complexities of real-world implementation. In the space of just a few months, they’ve moved from an initial blueprint to their first partnerships with financial and research institutions, and fast growing companies. This is what we love to see: technical brilliance mixed with level-headed pragmatism in solving customer problems at breakneck speed.

I’m proud to have Air Street support them as they begin on their journey.

You can read more about their work on the Fern blog here and over at Fortune from Jeremy Kahn.

The team is growing - if you’re interested in the sharp end of agentic AI implementation and work on infrastructure (both cloud and model training) and large scale systems engineering, check out open roles here.