Building the reasoning infrastructure for experimental science.

The most valuable scientific knowledge lives not in papers – but in the failed conditions, protocol changes, and expert judgment calls that happen at the bench. Labtree captures that layer.

Backed by
Liquid 2
Ada Ventures
With angels from
Google DeepMind
OpenAI
Snowflake
01

Experimental science remains largely invisible to AI. The most valuable knowledge at the bench lives in failed conditions, protocol changes, troubleshooting paths, and the judgment scientists use in real time to decide what to change, repeat, or discard. Almost none of that reasoning enters structured systems – which means critical context is lost and scientific knowledge fails to compound.

02

We believe this is one of the fundamental bottlenecks in AI for biology. Models can learn from papers and datasets, but they cannot learn how experiments are actually designed, adapted, debugged, and interpreted if the reasoning behind those decisions is never captured. The missing layer is not more computation. It is a reasoning substrate built from real scientific work.

03

Labtree is building that substrate. We capture reasoning traces from live experimental workflows, extract expert decision patterns that do not exist in literature or standard databases, and structure them into a proprietary corpus designed for retrieval, validation, and scientific reasoning. This creates the foundation for AI systems that can operate with greater relevance and depth inside real experimental environments.

04

Our long-term advantage comes from the data itself. Published knowledge is useful, but it does not contain the failure modes, contextual adjustments, and judgment calls that determine whether experiments work in practice. Those signals have to be captured directly from scientific behavior and expert review. That is the layer Labtree is building.

05

We are designing Labtree for a future in which experimental knowledge becomes durable, machine-readable, and compounding from the moment it is created. The teams that make scientific reasoning observable and reusable will define how AI operates in biology.