autoresearch

Autonomous research loops are orchestrated by separating fast experiment cycles from higher-level reflection, maintaining structured state tracking and literature notes to continuously update an experimental program. This allows research workflows to both optimize and discover.

Managing multi-hypothesis workflows requires workspace and state management that maintains research-state, findings, logs, and experiment records. This structured context tracking ensures research direction changes survive long sessions and repeated experimentation.

Autonomous research orchestration suits long-running, multi-hypothesis research workflows requiring repeated experimentation, reflection, and direction changes. It bootstraps literature and hypotheses before running iterative experiments and synthesizing results.

Progress reporting generates human-facing research presentations that summarize experiment trajectories, results, and next steps. This reporting operates alongside workspace state management to track findings and logs throughout the research lifecycle.

Domain routing directs execution to specialized skills for training, evaluation, inference, data processing, and paper writing. This allows the core orchestration loop to manage hypotheses while delegating technical execution to domain-specific tools.