Lifebit, a provider of secure biomedical data platforms and Trusted Research Environments (TREs), has launched AI Scientist Studio, announced in September at HumanX Amsterdam. The model-agnostic platform enables researchers to build and deploy custom AI scientist agents within secure Federated TREs, allowing organisations to use frontier AI models on governed health data while ensuring the data never leaves its source. Rather than requiring researchers to transfer data into external AI environments, Lifebit brings the AI to the data, so that analysis takes place where governed datasets already reside.

The problem the platform addresses

National health systems, biobanks and pharmaceutical companies operate under strict privacy and governance requirements that prevent patient-level data from moving to third-party AI scientist tools. Researchers are also constrained by pre-built tools that they cannot easily adapt or replace. Lifebit says AI Scientist Studio removes both barriers, allowing researchers to build their own AI scientists and bring the AI to the data, not the data to the AI.

How it works

Built on Lifebit’s Federated TRE architecture, the platform applies the Five Safes framework to agentic AI. Researchers can bring their own, or choose any, foundation model, scientific tool and knowledge base, and then build specialist agents for cohort discovery, deep phenotyping, analysis and evidence generation.

Agents execute on compute inside each data custodian’s governed environment, inheriting user permissions and operating federated across sites. Patient-level data never moves. Chat history, outputs and complete audit trails re-
main within the environment, and only approved results leave through Lifebit’s Airlock controls.

“AI Scientist tools are here, and they are the future of health data analysis, but the world’s most important health data still cannot use them”, said Dr Maria Dunford, CEO of Lifebit. “Researchers are constrained by pre-built tools, while existing AI Scientists require sensitive data to move outside governed environments. AI Scientist Studio changes this by bringing any AI model to the data, enabling researchers to build their own AI Scientists and run them securely, in a federated manner, across health and genomic data without compromising control, sovereignty or oversight.”

Dr Dunford noted: “As governments, health systems and life sciences organizations adopt frontier AI, they increasingly face the challenge of doing so without surrendering sovereignty over sensitive data and infrastructure. Major AI vendors have made significant moves into scientific research this year, but the prevailing model still assumes that data comes to the AI. Lifebit is inverting that model, bringing frontier AI directly to sensitive health and genomic data.”

Existing deployments

AI Scientist Studio builds on Lifebit’s federated secure infrastructure, which is already used by organisations including Genomics England, NIHR BioResource, Canpath, Singapore’s Ministry of Health and Boehringer Ingelheim. The platform supports the secure manage-ment and analysis of more than 275 million patient records across 30+ countries.

For more information, visit: https://lifebit.ai/
Digital issue: Please click here for more information