Our sister site:System Enablers — opens in a new tab

FHIL · ATLAS

Before you back the idea, test it with the people it must work for.

A proposal can explain the idea. It cannot show what happens when people try to use it.

ATLAS is FHIL's build-and-test engagement. We build a small first version of a health idea and test it with intended users or operators. We return the working form, the findings, a written recommendation, and a short work plan.

Two decision needs

Start with the decision the test must inform.

For foundations and programme leaders

You need to know whether a proposed way of working deserves support. The test produces evidence to inform a programme-funding request, a funding decision, or reinvestment in existing work.

For health investors

You need to know whether an operating assumption holds before committing more capital. The test examines how the proposed model works with the people and workflows it depends on. It informs that part of your diligence—not the entire investment decision.

How ATLAS works

Build. Measure. Learn.

01

Build

Agree the assumption that could change the decision. Make the smallest first form needed to test it. This may be software, a workflow, a service pathway, or a field tool.

02

Measure

Try that form with intended users or operators under agreed safeguards. Observe what they can do, where the pathway breaks, and what still requires human work.

03

Learn

Use the findings to refine the model and recommend what happens next: proceed, add conditions, redesign, test further, or stop.

The first form is a learning experiment. The purpose is to find out whether the idea deserves further support, not to secure approval for it.

The handover

What you receive

01

A working first form

The small version used in the test, which you can inspect and share.

02

A governance and data model

Who decides, what safeguards apply, and what information and data sources the pathway needs.

03

An evidence and decision record

What happened, what the findings support, what remains unknown, and FHIL's recommendation.

04

A short work plan

The next actions, dependencies, and questions to resolve.

Scope and timing

A defined test, not a fixed number of days.

We agree the scope, fee, milestones, and delivery date around your decision and the work needed to inform it. Access to users, information, and permissions must support the proposed test.

Your institution retains authority over current care, safeguards, and funding. Further implementation is scoped separately. You do not need to commission a Decision Brief before ATLAS.

Ownership, authority, and confidentiality

The client retains ownership of its data. FHIL retains its pre-existing methods, templates, tools, software components, and know-how, subject to the engagement agreement.

Clinical, legal, regulatory, ethics, security, procurement, data, funding, and institutional approval responsibilities remain with the relevant authorities.

FHIL does not identify the client or publish engagement material without written permission.

Evidence in the work

First application: ATLAS Rare Disease

A U.S. family foundation backing a rare-disease patient community commissioned a two-stage test of whether specialist knowledge could become a responsible AI-assisted patient pathway.

FHIL built a source-oriented knowledge base, a patient on-ramp and prompt system, and a distribution package. The technical core was feasible. The expected distribution channel and paying demand were not established. The buyer funded two stages and paused further work based on the findings.

Rare disease was the first application of ATLAS. It does not define the scope of the engagement.

Read the rare-disease case

A small test informs the next commitment; it does not establish clinical effectiveness or readiness for scale. The rare-disease case retains its full evidence boundaries in the case record.

Begin a conversation

Bring the idea and the decision ahead.

Tell us what you want to test, who it must work for, and what decision the findings need to inform. Include the date if one is set.

Discuss your idea

Do not include patient-identifiable, protected-health, or institution-confidential information in your initial email.