Labs

Things we build.

Prototypes, tools, frameworks, datasets.

Open and internal work from the lab. Tools that grew out of a research question, frameworks for understanding how intelligence is built, prototypes worth shipping for their own sake.

What lives here

Real artifacts. Things that run.

Open-source prototypes, datasets, and reference implementations from the lab’s research questions. Each entry is something you can clone, install, run, or download. Nothing on this surface is theoretical. The research framing for each release lives next to it as a research note.

Three categories.

01

Tools

Small prototypes that solve a specific substrate-layer problem. Released open-source when they work. Single-purpose by design.

02

Frameworks

Opinionated scaffolding for studying or operating on the foundational layer. Bigger than a tool, smaller than a product. Composable across research questions.

03

Datasets

Curated training-data subsets, evaluation benchmarks, reference corpora. Released when the methodology is reproducible and the documentation is honest about what is and is not measured.

Currently

Nothing released yet.

The first entries are in development. The lab ships when the methodology is reproducible and the documentation is honest about scope. Nothing leaves the lab as a marketing artifact; what gets released either teaches the field something or reduces friction for someone doing the work.

Until then, research notes and papers trace the questions that will produce the first artifacts on this surface.