Disarray builds AI research agents for long-horizon autonomous research over complex, heterogeneous proprietary data
Disarray builds AI research agents for long-horizon autonomous research over complex, heterogeneous proprietary data. Its agents manage the end-to-end experimentation loop, while a context graph links data, code, execution history, and documentation to help agents form stronger hypotheses and avoid failures in extended workflows. Its first product use case is autonomous machine-learning model development.