OmniScientist is an end-to-end, omni-modal AI scientist that conducts multidisciplinary research directly from heterogeneous raw evidence. It uses a perception layer and three autonomous agents for ideation, experiment, and writeup within a deterministic pipeline. The system enforces novelty screening, statistical validity, execution provenance, and numerical traceability via idea, rigour, and claim checks in code. It was evaluated on 36 real-data cases spanning 5 discipline families, 4 families of scientific evidence, and multiple modalities.
OmniScientist is an AI scientist designed to automate complete research workflows from hypothesis generation to manuscript preparation, while incorporating heterogeneous raw evidence such as spatial, temporal, cross-channel, and procedural relations. It employs a perception layer and three autonomous agents (ideation, experiment, writeup) in a deterministic pipeline, with code-based checks for novelty, rigor, and claim validity. The system was evaluated on 36 real-data cases across 5 discipline families and 4 evidence families.
The system's deterministic pipeline and code-based checks for novelty, statistical validity, execution provenance, and numerical traceability suggest a focus on reproducibility and verifiability in AI-driven research. The use of a perception layer to process heterogeneous raw evidence indicates multimodal input handling beyond text and code.
OmniScientist represents a step toward fully automated scientific discovery, potentially reducing the time and cost of research across multiple disciplines. Its emphasis on evidence-based reasoning may increase trust in AI-generated scientific outputs.
The system could accelerate R&D cycles in pharmaceuticals, materials science, and other evidence-intensive fields, offering potential cost savings and faster time-to-insight for organizations that adopt it.
Observable next signals include publication of detailed evaluation results, release of code or demos, and adoption by research institutions. Further development may expand the range of supported evidence modalities and disciplines.