Skip to content
MedeviRequest access

About Medevi

Open, validated AI for medical evidence synthesis.

Medevi is building the next generation of medical research infrastructure: powered by AI, validated by science, and open by design. Our mission is to accelerate evidence synthesis while keeping it rigorous, reproducible and transparent.

Our principles

Medevi at the intersection of rigorous science and machine learningTwo overlapping circles. Rigorous science: expert benchmarks, pre-registration, reporting standards. Machine learning: model-agnostic LLMs, uncertainty quantification, calibration. Medevi software sits where they overlap.Rigorous scienceExpert benchmarksPre-registrationReporting standardsMachine learningModel-agnosticUncertainty estimatesCalibrationMedeviSOFTWARE
Open by design

Model-agnostic and open source. Use any large language model and keep full control of your data and methods.

Benchmarked

Measured against expert human reviewers on public datasets, and we publish where models fall short.

See our research →
Interpretable

Every output shows its reasoning and the evidence it rests on, so experts can check it.

Reproducible

Standardized prompts, calibrated confidence and open code, so results can be rerun and verified.

Integrated

Built to fit the research and clinical tools people already use, not to replace them.

Medevi builds software where rigorous, evidence-based science meets modern machine learning.

Where we are

Medevi was developed at the Michael G. DeGroote School of Medicine, McMaster University, with institutional support from Microsoft, Cohere Labs, the Digital Research Alliance of Canada and McMaster University. The platform is in private alpha with research teams. Request access or meet the team.

Collaborate

We are a growing academic collective and looking for skilled computational collaborators. Reach us at team@medevi.ca if you would like to collaborate.