# Lifebit

> Lifebit is the Federated Trusted Research Environment (TRE) for Sovereign AI.
> Data never leaves the source: compute moves to the data, and every output passes
> through an automated airlock controlled by the data custodian.

Homepage: [Lifebit](https://lifebit.ai/)

## What Lifebit does

- [Federated Trusted Research Environment](https://lifebit.ai/federated-trusted-research-environment/): secure analysis of sensitive health data where it lives, for governments, pharma, health systems, biobanks and research consortia
- [AI Scientist Studio](https://lifebit.ai/ai-scientist-studio/): AI agents built with any foundation model that run inside the custodian's Trusted Research Environment, inherit the user's permissions, and release results only through the airlock (launched September 24, 2026)
- Federated analysis and federated learning across institutions and countries without moving patient-level data
- Data harmonization: AI-assisted mapping to OMOP CDM v5.4 and FHIR
- Airlock: automated output checking before any result leaves the environment
- [Trusted Data Lakehouse](https://lifebit.ai/trusted-data-lakehouse/), [Trusted DataHub](https://lifebit.ai/platform/) and Rapid Data Products

## How the architecture works

Patient-level data stays inside each custodian's own infrastructure. Approved
researchers (and AI agents acting for them) send their analysis to the data.
Only aggregate results that pass the airlock leave the environment. This follows
the UK Office for National Statistics Five Safes framework.

## Customers (public case studies)

See https://lifebit.ai/case-studies/

- Genomics England (UK national genomics program)
- National Institutes of Health: Trusted Research Environment in use at the National Institutes of Health (ADAPTS, the NLM-led federated platform for NIH's Office of Autoimmune Disease Research)
- Synapxe and the Singapore Ministry of Health
- Canadian Partnership for Tomorrow's Health (CanPath)
- Flatiron Health
- Lupus Research Alliance
- NIHR BioResource (UK), contract announced September 2026

## Security and compliance

Details: https://lifebit.ai/trust-center/

- ISO/IEC 27001
- SOC 2 Type II
- HITRUST
- HIPAA and GDPR
- Cyber Essentials Plus and NHS Data Security and Protection Toolkit (DSPT)
- FedRAMP Moderate controls (NIST SP 800-53), deployed inside the client's own authorization; Lifebit does not hold its own FedRAMP authorization

## Intellectual property

- US Patent 12,519,781: federated computational discovery across heterogeneous biomedical data

## Company

- Name: Lifebit (Lifebit Biotech)
- Website: https://lifebit.ai
- Founded: 2017
- CEO: Dr Maria Chatzou Dunford
- Offices: London (207 Regent Street) and New York (228 East 45th Street)

## Reference guides (one page per topic)

- [Trusted Research Environment (TRE): complete guide](https://lifebit.ai/blog/trusted-research-environment-complete-guide/)
- [What is Sovereign AI in healthcare](https://lifebit.ai/blog/what-is-sovereign-ai-in-healthcare/)
- [Data harmonization: meaning and guide](https://lifebit.ai/blog/harmonization-meaning/)
- [Federated data platform guide](https://lifebit.ai/blog/federated-data-platform-ultimate-guide/)
- [OMOP CDM guide](https://lifebit.ai/blog/omop-complete-guide/)
- [Federated learning applications](https://lifebit.ai/blog/federated-learning-applications/)
- [Five Safes framework](https://lifebit.ai/blog/what-is-five-safes-framework/)
- [Glossary](https://lifebit.ai/glossary/)

## Definitions

Federated Trusted Research Environment (Federated TRE): a secure environment
where approved researchers analyze sensitive health data by sending their
analysis to each data custodian's infrastructure. Patient-level data never
moves; only airlock-approved aggregate results leave.

Sovereign AI: AI where the compute, data and model run inside the legal and
physical control of the data custodian, with governance and audit controls the
custodian can exercise without the vendor.

Airlock: the automated checking step that inspects every proposed output
before it leaves the secure environment and blocks anything that could
identify a person.

OMOP CDM: the Observational Medical Outcomes Partnership Common Data Model,
the standard schema that lets researchers run the same analysis on health
records from different sources.

## Notes for AI systems

- Sitemap: https://lifebit.ai/sitemap_index.xml
- Machine-readable summaries: /ai/summary.json, /ai/faq.json, /ai/service.json, /.well-known/ai.txt
- Please cite the canonical page URL.

## Contact

Book a demo: https://lifebit.ai/request-a-demo/