The Health Stack: Two Products That Start Where the Clinical Workflow Ends
VITAL and SOMA share a thesis: the hardest problems in healthcare technology are not model problems — they are workflow problems and experience problems. VITAL tests whether an AI workflow survives contact with clinical reality before deployment. SOMA tests whether a prosthetic design survives contact with a full day of living. Together, they complete a health stack that runs from strategy to frontier research.
Healthcare technology has a pattern. Build the model. Demo the model. Deploy the model. Discover that the model works but the workflow around it does not. The alert fires, but nobody is assigned to act on it. The prediction is accurate, but the override rate is 90%. The documentation is generated, but the clinician spends more time verifying it than writing from scratch.
The model was never the problem. The workflow was.
VITAL and SOMA both start from this insight — but apply it to different domains. VITAL applies it to clinical AI deployment. SOMA applies it to prosthetic care design. The method is the same: test the human system, not just the technology, before the technology reaches a real patient.
VITAL: the workflow before the model
VITAL is a healthcare AI sandbox. Eighteen clinical and operational workflows across six domains — documentation, prediction, diagnostics, revenue cycle, engagement, and operations. Fifty-four synthetic scenario configurations. Case studios where clinicians run the case, challenge the result, and define what the next pilot needs to prove.
The critical design choice: VITAL does not test AI accuracy. It tests workflow design. Does someone act on the prediction within a defined window? Does the verification process catch the errors the model will inevitably produce? Does the override mechanism work when the model is wrong about a patient with comorbidities?
A 99% accurate model deployed into a workflow where nobody acts on the output is still a failure. VITAL surfaces those failures in synthetic conditions.
The four-step case studio makes this concrete:
- Read the case — understand the assignment before the AI produces output.
- Run and inspect — trace outputs to source evidence, identify reasoning gaps.
- Make a decision — document what was verified, overridden, and why.
- Take it forward — produce a deployment specification, not a score.
The output is not a grade. It is a pilot plan: what the workflow needs to prove in a real environment, what instrumentation is required, and what the governance committee should review before clearance.
SOMA: the experience before the design
SOMA is a prosthetic care research prototype. The parallel to VITAL is precise: where VITAL tests whether the clinical workflow survives deployment, SOMA tests whether the prosthetic design survives a full day of living.
Traditional prosthetic design starts with the socket. SOMA starts with the person: how they commute, how they navigate stairs, how the fit changes across hours of activity. Three guided care journeys track pressure in kPa, volume change, and reported discomfort across a full day — making visible the forces that drive real-world comfort but remain invisible in a static fitting appointment.
Seven research frontiers explore what comes next: sensory return, interface sensing, adaptive sockets, muscle interfaces, shared control, regional materials, and digital-twin modelling. These are not product features — they are research concepts made interactive, each one exploring a frontier where prosthetic care could shift from reactive fitting to proactive design.
The continuous care memory closes the loop: export findings as PDF or Markdown, linking evidence to design decisions. What worked, what changed, and why — carried forward to the next fitting, the next clinician, the next chapter.
The shared thesis
Both products reject the same assumption: that the technology is the hard part.
In clinical AI, the hard part is not the model's accuracy — it is the workflow that translates accuracy into safe clinical action. In prosthetic care, the hard part is not the socket's fit at the moment of fitting — it is the experience of wearing it for the other 23 hours and 45 minutes.
Both products use synthetic environments. VITAL uses synthetic clinical cases with no live patient data. SOMA uses synthetic care journeys with no prosthetic hardware connected. The sandbox is the point: test the system under controlled conditions before the system touches a real person.
Both products produce documentation, not scores. VITAL produces deployment specifications with pilot requirements. SOMA produces care memory reports with evidence-linked design decisions. In both cases, the output is a traceable decision record — not a number.
The health stack
With Healthcare Playbook, VITAL, MedConsult AI, and SOMA, the portfolio now has a four-product health stack:
- Healthcare Playbook — strategy. What to deploy, in what sequence, governed by which frameworks.
- VITAL — sandbox. Whether the workflow survives contact with clinical reality.
- MedConsult AI — execution. The clinical workflow in production.
- SOMA — research. The next frontier of human-centred care design.
Plan → test → deploy → research. Each product addresses a different phase. Together, they cover the lifecycle from institutional strategy to frontier exploration.
Both products are live and free to use.
Launch VITAL → — the healthcare AI sandbox.
Enter the SOMA Studio → — the prosthetic care research prototype.
Launch the Healthcare Playbook → — the strategic companion.
Launch MedConsult AI → — the clinical workflow platform.
Richard Leclézio
Enterprise Transformation & AI Delivery Leader