Happy August — and Happy GME Professionals Day.
Today is an opportunity to recognize the Program Directors, Program Coordinators, DIOs, faculty leaders, and other GME professionals whose work keeps residency programs moving every day. Their expertise, judgment, and persistence are essential—and much of what we are building is intended to give some of that time and attention back.
This month, the Family Medicine pilot is coming into focus around five connected ideas: Run Alongside, In the Moment, the Curated FM Intelligence Layer, Entrustment Everywhere, and Time Savings & More.
Together, they describe both how we intend to test GME Manager and what we want to learn: can we add meaningful AI-supported workflows alongside the systems programs already use, capture better evidence closer to the clinical moment, preserve faculty judgment, and measurably reduce the work of residency?
─────────────────────────────────
1. Run Alongside
Programs should not have to replace an entire platform just to prove that a better workflow works.
That is the idea behind Run Alongside. GME Manager works alongside New Innovations, MedHub, or another existing system of record, performs selected residency workflows, and returns the completed, human-reviewed output to the environment the program already uses.
For the Family Medicine pilot, we are starting with two core workflows: Procedure Logging and Evaluations.
With Procedure Logging, residents can capture the procedure close to the clinical encounter, let AI prepare the structured log, review and edit it, and send it to faculty for approval. Faculty retain responsibility for reviewing the evidence, confirming the appropriate supervision or entrustment judgment when applicable, and approving the final record.
With Evaluations, faculty can capture observations close to the encounter, preserve them as private evidence, and later select the evidence they want AI to use in preparing an evaluation draft. Faculty retain responsibility for the final ratings and entrustment judgment, reviewing and editing the narrative, and submitting the completed evaluation.
Run alongside first. Prove the workflow. Expand from what works.
Programs do not need to migrate data, replace their current platform, or redesign their operating model simply to test a better way of working. That lower implementation lift can enable current and future pilots to go live in days, measure results against a defined baseline, and expand based on evidence rather than promises.
─────────────────────────────────
2. In the Moment
Sometimes documenting the work takes longer than performing it. That should be unacceptable. Starting today, we are treating it that way.
“In the Moment” is the operating idea behind the pilot—and it was inspired by and co-developed with one of our pilot sites. The idea is simple: capture procedure details and faculty observations during or immediately after the clinical encounter—not hours or days later, when the details have to be reconstructed from memory.
For Procedure Logging, a resident can capture the procedure and supervision details while the experience is still fresh, then review and confirm an AI-drafted log. For Evaluations, a faculty member can capture a brief observation in seconds and later choose that observation as source evidence for an end-of-rotation evaluation.
Capture once, close to the work, and build from that capture. The goal is faster documentation and better evidence—less reconstruction from memory and a clearer chain from the clinical moment to the educational record.
We are also designing around the interruptions that define clinical work. Once a moment is captured, it should remain available as a private draft so residents and faculty can return later to review, edit, delete, or complete the workflow rather than recreate the event from memory.
─────────────────────────────────
3. Curated FM Intelligence Layer
A voice note without intelligence is transcription. A voice note with intelligence becomes evidence.
More precisely, the Curated FM Intelligence Layer gives a captured moment the specialty-specific context needed to interpret it, structure it, and prepare it for human review and reuse across GME workflows.
The current Family Medicine layer contains 15 pre-loaded data sets across three intelligence domains:
Foundational: ABFM procedures, ABFM optional procedures, procedure complications, ACGME milestones, the 14 ABFM core competencies, and duty-hour rules.
Clinical Intelligence: 218 Family Medicine diagnoses mapped to ICD-10, USPSTF screening guidelines, ACIP immunization schedules, behavioral health screeners, rotation objectives, CPT/E&M + G2211/APCM, and MIPS quality measures.
Documentation Conventions: Family Medicine note conventions and differential frameworks.
These are not static reference files sitting beside the workflow. They work inside it. The intelligence layer helps interpret a spoken procedure, apply supervision logic, connect observations to competencies and milestones, structure evaluation drafts, and organize evidence for CCC review and attestation.
Configure once. Apply the intelligence everywhere.
The same specialty context that helps create a procedure log should also make the evaluation, CCC evidence view, and eventual attestation more coherent. That consistency is what allows one captured moment to become reusable evidence across the residency record instead of another isolated form.
─────────────────────────────────
4. Entrustment Everywhere
Faculty make entrustment judgments every day. Too often, those judgments disappear instead of becoming durable evidence of resident growth.
Entrustment Everywhere grew directly out of pilot-site feedback. The concept was inspired by and co-developed with one of our pilot sites, which owns the concept. GME Manager is implementing it within the pilot as a way to make an important faculty judgment more visible, consistent, and reusable across the residency record.
The idea is straightforward: one five-level Family Medicine supervision judgment, used wherever faculty already make related decisions. In the pilot, that means three existing workflows—procedure approval, direct observation, and end-of-rotation evaluation. It is not a separate EPA system, and it is not an AI-generated competence score.
The safeguards matter. “Not observed” is not a level. Approving a procedure log and judging independence remain separate decisions. AI may organize the evidence or prepare language, but faculty own the entrustment judgment.
One scale. Three workflows. A longitudinal view of development.
If the approach works, individual faculty judgments can accumulate across procedures, observations, and rotations to create a clearer picture of resident development and readiness over time—without asking faculty to learn another evaluation system or complete another disconnected form.
─────────────────────────────────
5. Time Savings & More
There are two opportunities here: give residents and faculty time back by reducing administrative work, and give Program Directors, Program Coordinators, and CCCs better evidence to understand resident development and readiness.
The pilot is designed to answer a broader question than simply, “Is it faster?” We will evaluate success across three outcome categories: Time Savings, Evidence Volume & Speed, and Evidence Quality.
Time Savings means less time spent completing procedure logs and evaluations, less administrative work for residents and faculty, and less follow-up and rework required to get the work finished.
Evidence Volume & Speed means more logged procedures, evaluations, and entrustment judgments; faster completion and turnaround; and more evidence captured close to the clinical moment.
Evidence Quality means more complete, specific, and usable documentation; more meaningful evidence for Program Directors, Program Coordinators, and CCC review; and clearer longitudinal visibility into resident performance and readiness.
Save time. Capture more evidence. Make the evidence better.
That is the standard we want the pilot to meet. If we are successful, we will have proven something bigger: the administrative work of residency can get dramatically easier while the evidence behind its most important decisions gets better.
─────────────────────────────────
Coming This Fall: National Resident and Faculty Surveys
Later this fall—likely in September or October—GME Manager will launch national surveys of residents and residency faculty to build a broader, direct view of administrative burden, evaluation workflows, procedure logging, supervision documentation, CCC preparation, and perspectives on AI-enabled support in residency training.
These are intended as national listening efforts, not product surveys. We will share summary findings with the GME community after the surveys close and are analyzed. Programs and organizations that help distribute the surveys will also receive the more detailed findings, including the underlying de-identified response data.
We are looking for volunteers to help us reach residents and faculty across specialties, program types, and regions. Program Directors, Program Coordinators, DIOs, faculty leaders, chief residents, and others willing to forward one or both surveys can make an important contribution.
Interested in helping? Contact Michael at [email protected], and we will send a short, ready-to-forward message and survey link when distribution begins.
─────────────────────────────────
Interested in a Pilot?
If your program wants to be considered for a future pilot opening, raise your hand. The model remains simple: run alongside the platform you already use, capture evidence closer to the work, apply specialty-specific intelligence, preserve faculty judgment, and measure whether the result saves time while improving evidence.
Request a spot → [email protected]
─────────────────────────────────
Know a Family Medicine PD who should see this?
Please forward this issue. The more programs that challenge the ideas, volunteer for research, and tell us what does not work, the better GME Manager will become.
With gratitude,
— Michael Sousa
Founder & CEO, GME Manager
[email protected] | 720-352-5203 | gmemanager.ai
─────────────────────────────────
Follow GME Manager on LinkedIn.
Pilot progress, product decisions, and the honest week-to-week of building GME Manager will be shared there.