GPe Research GPe Research
AI is making decisions and no one is accountable   ·   THE ACCOUNTABILITY GAP (TAG™)   ·   the gap closes when both seats are named   ·   MedicoVigilance™   ·   a decision is only as good as your ability to answer for it later   ·   GPe Alpha   ·   AI is making decisions and no one is accountable   ·   THE ACCOUNTABILITY GAP (TAG™)   ·   the gap closes when both seats are named   ·   MedicoVigilance™   ·   a decision is only as good as your ability to answer for it later   ·   GPe Alpha   ·   
GPe Research

AI is making decisions. Most institutions cannot tell you who owns them.

The Architecture of Accountability.

That gap has a name: The Accountability Gap (TAG™). It lives at a specific moment in every regulated workflow. The moment the AI stops and the professional starts. GPe is the accountability infrastructure built for that moment.

SEE THE CATEGORY FOR INSTITUTIONS
TRAINING & INSTITUTIONAL PEDIGREE
Brown University MIT Columbia Business School London School of Economics IE Business School University of Minnesota
The Category
GPe is the accountability infrastructure for the decisions AI is making in regulated industries.

AI produces outputs. Institutions make decisions. GPe builds the architecture between those moments.

Not the model. Not the data pipeline. The named accountability architecture that sits between AI output and institutional decision. The governance owner who approved the system. The decision owner who made the call. The documented handoff between them that holds when a regulator or a board asks who owns this.

Every regulated industry where AI is making decisions needs this architecture. Most have not built it. GPe is the standard for what it looks like when they do.

AI does not think. It predicts. GPe makes the decisions those predictions drive hold.
The Opening

The moment the AI stops and the professional starts.

Three things happen. One is recorded. Two are not.

01 · RECORDED

A professional acted

Their name is on the chart, the file, the record. That much the institution has.

02 · NOT RECORDED

What the AI shaped

What the AI recommended, what the professional did with that recommendation, what shaped the call beneath it. None of that lives on the record.

03 · NOT RECORDED

Who owns the AI

Who approved the AI for use in this workflow, on this institution's authority, with what review. Not on the record either.

The Evidence

The Accountability Gap is not a theory. It is already measured, and already priced.

22.2%
Potential for severe harm from a model's medical recommendation, across 31 large language models. 76.6% of those errors were errors of omission.
NOHARM Benchmark · ARISE Network · Jan 2026
~70%
Top-model first-attempt success on Stanford's MedAgentBench. The best clinical AI agents are wrong on close to one task in three.
Stanford MedAgentBench · 2025
22%
Of US hospital leaders are highly confident they could produce a complete, auditable AI explanation for regulators within 30 days. Most could not.
Black Book Research · n=182 · Late 2025
CG 40 47
The ISO Commercial General Liability endorsement that, effective Jan 1 2026, excludes injury arising out of generative AI. The underwriters have already priced the risk.
ISO Form CG 40 47 · W. R. Berkley PC 51380
The Answer

Five pieces. Built to operate as one.

THE STANDARD

GPe Alpha

THE DISCIPLINE

MedicoVigilance™

THE CREDENTIAL

MVP Certification

THE ACADEMIC BODY

Journal of MedicoVigilance

THE RECORD

GPe Research Publications

THE STANDARD

GPe Alpha

GPe Alpha is a reliability score for a decision made under uncertainty. Probability tells you what is likely. GPe Alpha tells you how much the evidence behind the decision agrees with itself. Two readings on the same decision: one says it is probable, the other says the inputs hold together, or they do not. Read together, that is what makes the decision hold.

GPe Alpha /alpha a standard in development
THE DISCIPLINE

MedicoVigilance™

MedicoVigilance™ is the discipline of overseeing clinical AI in practice: watching deployed systems, catching harm before it spreads, and keeping the record of what each system did and who answered for it.

MedicoVigilance™ MVP certification Journal of MedicoVigilance MedicoVigilance™ newsletter
THE FRAMEWORKS

The frameworks, published

The frameworks name the owners, map The Accountability Gap (TAG™), and price the exposure.

Named Owner Mind the 9 Blocks™ Clinical AI Accountability Canvas™ Gap Score™ TAG™
The Named Owners

Two seats. One handoff. Filled by name.

01 · SEAT ONE

The Named Governance Owner

Answers when the question is: who approved this AI for use here?

Charter. The institutional decision to deploy.
Commission. The ongoing authority to keep it running or pull it.
Cover. The organizational accountability when the system produces harm.

02 · SEAT TWO

The Named Decision Owner

Answers when the question is: who made this call?

Decide. Receives the AI output and makes the call.
Document. Records what the AI recommended, what they decided, and why.
Defend. Stands behind the call in audit, in review, in front of the board.

03 · THE HANDOFF

Between them

The Governance Owner sets the conditions under which the AI operates. The Decision Owner acts within those conditions on a specific case.

The handoff between them is the moment the gap opens, and the only place it can close.

Research & Labs

GPe runs as two arms. One defines. One builds.

GPe RESEARCH

The standards arm

  • Owns and develops GPe Alpha, the standard for decision reliability
  • Authors the disciplines that adopt those standards, MedicoVigilance™ first
  • Publishes the institutional record: the Journal, the citation publication, the newsletter
GPe LABS

The build arm

  • Builds the products that put GPe Research's frameworks to work
  • Tests them in real clinical settings
  • Deploys with pilot partners, starting in health systems
The Aperture

The same gap opens wherever AI is making decisions. GPe starts where it is sharpest.

01

Healthcare — the deep vertical, active now

A model shapes a recommendation, a clinician acts, and a patient lives with the result. The accountability is bodily, the regulatory architecture is mature, and the GPe stack maps cleanly onto what is already there.

MedicoVigilance™ Named Owner Mind the 9 Blocks™
02

Finance, insurance, and law — the same break, under audit

An AI output shapes a decision a regulator can later question, and the institution has to show the evidence, the reasoning, and the owner. The break is identical; only the body of law changes. The standard travels first. The discipline names follow.

03

Every regulated industry — one standard, horizontally

Anywhere AI is making decisions faster than anyone can answer for them, the same gap opens. GPe Alpha is built to travel: a single reliability standard that any sector can adopt.

The Proof

Published, cited, and available today.

FRAMEWORK BRIEF Ñ01

Mind the 9 Blocks™: a diagnostic canvas for clinical AI accountability

A nine-block canvas and a three-part Gap Score™ for surfacing, scoring, and closing the clinical AI accountability gap before a recommendation reaches a patient.

Mo Johnson, MD MBA · May 22, 2026 · 24 min read
POSITION PAPER Ñ01

Named owner: why every clinical AI output needs a professional on record

The full case for the Named Owners architecture, with citations in five formats. The published paper behind this page's argument.

Mo Johnson, MD MBA · 9 min read

The full record

The full record. Working papers, framework briefs, position papers. Citations in five formats, indexed and readable today.

VISIT PUBLICATIONS
The Stack

GPe sits above the data and AI stack, where the decision is made.

Every current system stops one step before the accountable decision. GPe is what closes that step.

GPe — ACCOUNTABILITY
The accountable, audited decision, with a name attached. The standard, the discipline, the credential, the record. The gap, closed.
AI MODELS
Generate AI output. They do not own it.
DECISION PLATFORMS
Advise and surface. They do not name who owns the decision.
DATA PLATFORMS
Process and store the data. They do not name what was decided with it.
For Institutions

If you carry the deployment, you are already on the record. The next step is making it hold.

For the executives, counsel, and boards already named when an AI-shaped decision is questioned. Three ways to bring the accountability standard into your institution.

Brief your governance committee

A focused briefing on the Named Owner Principle and the Mind the 9 Blocks™ canvas, scoped to your deployment landscape.

REQUEST A BRIEFING

Get the framework

The full Mind the 9 Blocks™ canvas and Gap Score™ methodology, published with citations in five formats. Bring it to your next governance review.

READ THE FRAMEWORK

Audit a live deployment

Apply the Gap Score™ to a clinical AI deployment already in your workflow. Surface the Named Owner gaps while you can still close them.

CONTACT RESEARCH
About

GPe builds the accountability that AI cannot.

GPe is a standards body and research practice for regulated work. It defines the standard, shapes the discipline, credentials the professional, and publishes the record. Together, they make the decisions AI is making hold.

A decision is only as good as your ability to answer for it later, with the evidence, the reasoning, and the name of who owned it.

Healthcare is where it begins, because the stakes are bodily and the architecture is mature. Every other regulated sector follows the same line. AI is making a decision, a professional puts a name to it, and the institution has to answer when asked. GPe is the work of making those answers hold.