Turn AI principles into requirements institutions can actually demonstrate.
AI is starting to help with decisions about people's health, in hospitals, and in pro sports, where teams use data about players' bodies to decide who plays and when. Darwin by Medigram is the system that checks the AI: it scores every decision on six kinds of trustworthiness and keeps a permanent record, so whether the stakes are a patient's care or a player's career, nothing happens unchecked and there's always proof. Designed to work on mobile, where the work happens.
Here is the plain version: high-level principles matter, but institutions ultimately need requirements that engineers can implement, clinicians and executives can operate under, security teams can test, and organizations can demonstrate with evidence.
A short reader path into Building Darwin, the full CEO Letter.
The implementation gap
The gap appears when policy discussions operate at a different level from the systems that must carry them out. Accountability becomes meaningful operationally only when authority, boundaries, escalation, verification, and evidence can be expressed in the architecture itself, not only in a document.
Accredited standards provide a mechanism for converting broad objectives into durable requirements that diverse institutions, technical teams, and other stakeholders can use. Implementation then tests whether those requirements are precise and workable enough to survive contact with a real system.
Why organizations get this wrong
Medicine, engineering, cybersecurity, research, governance, and operations each use different methods for establishing trust. A requirement can be sensible within one discipline and still fail at the handoff to another. Someone has to close those seams.
Keeping the handoff from losing information
A principle written at the policy level and a control an engineer can actually build are not automatically the same thing, and the gap between them is where good intentions quietly stop working. This architecture exists to keep that handoff from losing information: one path from principle, to accredited requirement, to working implementation, to independent verification, to evidence.
Why Sherri and the Medigram team
That path holds together because Sherri Douville’s own work crosses accredited standards authorship, architecture, AI engineering, security, behavioral verification, evidence design, and implementation, so a requirement does not have to be reinterpreted fresh at every handoff. The objective is to institutionalize that same method, so turning a principle into a demonstrable, testable requirement becomes repeatable infrastructure rather than something that depends on one person’s range.
What policymakers can look for
Can institutions translate principles into named authority and technical controls? Can independent parties test those controls? Are failures visible? Is remediation traceable? Can the institution demonstrate what occurred during a consequential decision?
Need the full architecture and evidence?
The CEO Letter contains the detailed technical rationale, verification loop, standards context and diligence path.