Showcase what's possible in primary care.
You already know what's broken. Share with everyone how you'd fix it.
This is a learning morning. You'll spend it building with AI tools — some of you for the first time — alongside colleagues who have already done it and can show you how. The prizes are real, but the point is to stretch: take a problem you've quietly wished someone would fix, and find out how far you can get with it before lunch.
Bring a problem you actually care about
Not the easiest one to solve — the one that's been bothering you for years. Think harder about why it's still broken than about what to build, and the building gets easier.
Learn the tools by using them
You don't need to write code. AI build tools have collapsed the distance between describing an idea and having a working version of it, and this is a supervised first run at that.
Colleagues who've already done it
Physician mentors on the floor all morning. They've built things, hit the walls you're about to hit, and can get you unstuck in two minutes rather than twenty.
Choose your side of the exam room.
Patient Empowerment
Help your patients take a more active, informed, and confident role in their own care.
Read the brief →Clinical Team Empowerment
Get your own day back — less time fighting the system, more time practicing medicine.
Read the brief →How it runs.
Welcome & Keynote with Dr. Mui
Doors, coffee, and the opening keynote — why this morning exists and what the two tracks are asking of you.
Team Formation & Problem Selection
An icebreaker to find teammates and start ideating on problems you'd like to solve.
Hackathon Work Session & Design Iteration
Two and a half hours of building, with mentors circulating the whole time.
Presentation Submission
Record and submit your three-minute video.
Awards & Closing Remarks
Winners announced at the Main Stage Event, Karl Dean Ballroom.
Two tracks.
Pick one track during problem selection and stay in it — scoring happens within tracks, and there's a winner in each. Both briefs are deliberately wide.
Patient Empowerment
Patients are often expected to navigate complex medical information, fragmented systems, difficult decisions, and unclear next steps with limited support. This track challenges teams to design solutions that give patients greater knowledge, agency, access, and control throughout their healthcare journey.
Teams may address challenges such as
- Understanding diagnoses, test results, medications, or treatment options
- Preparing for appointments and asking better questions
- Navigating referrals, insurance, costs, and follow-up care
- Managing chronic conditions between visits
- Accessing trustworthy and understandable health information
- Communicating needs, goals, preferences, and symptoms
- Reducing barriers related to language, health literacy, disability, or technology
- Helping caregivers and families participate appropriately in care
Clinical Team Empowerment
Physicians and clinical teams regularly encounter administrative burden, fragmented information, inefficient workflows, communication failures, and repetitive manual work. This track challenges teams to create solutions that help healthcare professionals work more effectively, collaborate more easily, and focus their attention on the work that matters most.
Teams may address challenges such as
- Reducing documentation and administrative work
- Improving clinical decision support without creating alert fatigue
- Making relevant patient information easier to find and act upon
- Improving communication between clinicians, staff, patients, and outside organizations
- Simplifying referrals, prior authorizations, results management, or care coordination
- Supporting quality improvement, peer learning, or clinical feedback
- Reducing cognitive overload, burnout, and unnecessary interruptions
- Improving team-based care and ensuring work reaches the right person
- Helping clinicians identify and address gaps in care
Plenty of good ideas serve both your patients and your team. Declare the track whose primary user you're designing for — the one whose day gets measurably better — and weigh your tradeoffs from there. If you're genuinely torn, pick the one where you can point to a specific person you know.
From a problem to a clickable prototype.
Answer a few questions about the problem you're solving. This writes the prompt for you — paste it into whichever AI tool you're using and you'll get back a single page you can click through on camera.
Just enough direction to start experimenting.
The prompt is built around the three-minute video structure, so the screens line up with what you'll need to say — and you get a timestamped narration script back alongside the prototype.
Four hours, paced.
The five blocks below are the official run of the hackathon. Inside the long work session we've suggested a rhythm — those checkpoints are guidance you can bend to your project.
Welcome & Keynote with Dr. Mui Keynote
Doors, coffee, and the opening keynote. Why this morning exists, what the two tracks are asking of you, and how the judging works. Arrive on time — the framing here is what separates teams that pick a good problem from teams that spend an hour choosing.
Team Formation & Problem Selection
An icebreaker activity to help you find teammates and start ideating on problems you'd like to solve.
Hackathon Work Session & Design Iteration Build
The main event. Mentors circulate the entire block — flag one down the moment you're stuck rather than burning twenty minutes on it. There are no scheduled interruptions, so pace yourselves against the checkpoints below.
- 9:30First mentor check. Pitch your problem, not your solution, and let a mentor stress-test whether it's narrow enough.
- 10:00Ugly version of your core interaction working end to end. If it isn't, cut scope now rather than at 11:00.
- 10:30Second mentor check. Show what you have, however rough. Last good moment to change direction.
- 11:00Feature freeze. Stop building. Decide exactly what the video will show and in what order.
- 11:15Start putting final touches on the concept.
Presentation Submission Closes 12:00
Record your three-minute video and submit it with any supplemental files. Half an hour is enough for two or three takes if you've already decided what you're showing — and not nearly enough if you haven't.
Check your sharing settings before you submit. A link that won't open can't be reviewed. See the submission instructions for the full checklist.
Judging
Enjoy the conference while your submissions are being evaluated.
Awards & Closing Remarks Main Stage
Winners announced at the Main Stage Event on Thursday, October 22, in the Karl Dean Ballroom — both track prizes plus the "OMG This Is So Cool" Award.
Submit a three-minute video.
Every team submits one recorded video overview of what they built, with a shareable link, plus any supplemental files. The submission window is 11:30 – 12:00.
One entry per team. You can edit your answers until the form closes at 12:00.
Form not working? Email your submission to [email protected] instead — same deadline, same details.
Use a free screen-recording tool.
Both of these are free, run in the browser, need no editing skills, and give you a shareable link the moment you stop recording. If you already have a tool you like, use that instead — we only care that the link opens.
Loom →
Records your screen and your face at once. Generous free tier, instant link, and viewers don't need an account. The safe default if you've never done this.
Tella →
Slightly more polished output — clean backgrounds, easy trimming, and you can re-record a single section without starting over. Worth it if you want to take a second take.
Narrate over whatever you built while you click through it, or narrate over a short slide deck — both work, and neither is penalised. Scoring is about how clearly the concept lands, not how finished it looks. Just make sure there's something on screen rather than only a talking head.
What to cover in three minutes.
Not a rule, but it helps guide you to include the essential information in the right sequence.
The problem
One specific clinical moment that goes badly today. Name the patient or the clinician. If a colleague wouldn't nod within fifteen seconds, it's too abstract.
The concept
The bulk of your time. Walk through what you built — prototype, mockup, or slides — and narrate how it would work. Show the one core interaction rather than touring every screen.
Adoption
Who it's for, who would champion it inside a practice, and how it realistically gets picked up — including who would pay for it and out of which budget.
Barriers and what's next
What would stop this from being adopted if it existed. Say plainly which parts are illustrative. Then one sentence on what you'd do next.
What the form asks for.
Eight questions, in this order. Only one of them takes any real thinking — draft question 5 before 11:30 and the rest is typing.
Your email
Whoever is submitting on behalf of the team. One entry per team.
Team name
Anything you like. Pick it before 11:30 so you're not naming your team at 11:58.
Team member names
Everyone on the team — prizes follow this list. The first person you list is the point of contact, and you'll need their mobile number, so have it ready.
Your track
Patient Empowerment or Clinical Team Empowerment. One only — you're scored against the other teams in that track.
Problem, who benefits, and your solution
Three questions in one box, answered in two or three sentences total:
- What problem are you solving?
- Who benefits from your solution?
- What is your solution?
Which tools you used
A checklist of the common AI build tools, with an "Other" box. Tick everything you touched — using AI heavily is expected here, not penalised.
Link to your video
A shareable Loom, Tella, or equivalent URL, set so that anyone with the link can view it. Three minutes maximum.
Link to any other files
Your prototype, slide deck, screenshots, workflow diagrams — anything that helps convey what you built.
Easiest approach: drop everything into one Google Drive folder, set it to "anyone with the link", and submit that single link. These are reviewed but never a substitute for the video.
The form asks for a video link, and optionally a link to your files. Before you paste either one in, open it in a private or incognito browser window — a window where you are not signed in to anything.
If it asks you to sign in, or says you need to request access, we can't open it either. A link set to "only people in my organization" is the single most common way to lose a morning's work.
Links checked in an incognito window, video under three minutes, everyone on your team named.
Form not working? Email your submission to [email protected] instead — same deadline, same details.
Scoring rubric.
Five criteria, 20 points each, 100 total.
You have two and a half hours. Nobody expects a finished product, and scoring is not about how sophisticated the build is. What matters is whether the concept comes across clearly — what the problem is, who it's for, and whether you've thought about what it would take for it to exist. A rough prototype with a sharp, well-reasoned idea beats a slick one you can't explain.
The five criteria
| Criterion | Points | What earns points |
|---|---|---|
| Clinical problem insightWould a colleague recognize this problem in the first fifteen seconds? | 20 | A specific patient or a specific clinician in a specific moment — not a category like "care coordination." Evidence the team understands why the problem has survived this long and why the current workaround fails. This is the criterion you are uniquely positioned to win, and the one most teams underinvest in. |
| Solution quality & creativityIs this a good answer to the problem you just named — and is it clear who it's for? | 20 | The idea follows directly from the problem rather than arriving from somewhere else, and the team is specific about who would actually use it. Thoughtful design decisions and an approach that goes beyond wrapping a chatbot around an existing task. Novelty counts only in service of the problem. |
| Workflow & real-world fitHas the team thought about what it would take for this to actually exist? | 20 | Where it lives in the day, who operates it, and what it costs in attention. How it might get paid for, and whether it attaches to a budget someone already has. An honest account of the barriers to adoption — IT, trust, training, regulation, inertia. Naming a hard barrier scores better than not mentioning it. |
| Prototype & video executionCould someone explain your idea back to you after watching once? | 20 | Clarity, not sophistication. Something visual on screen that makes the concept concrete — a working prototype, a clickable mockup, or well-built slides all count equally here. Clear narration, sensible pacing, finishes inside three minutes. Explicit about which parts are real and which are illustrative; candor scores, overclaiming does not. What two and a half hours buys is well understood. |
| Impact potentialIf this existed, who is meaningfully better off? | 20 | A clear answer to who benefits most, and why them rather than someone else. Plausible sense of how far it could reach beyond one practice, and a credible next step past the hackathon. Extra weight when the team pressure-tested the idea with a mentor or a colleague during the morning and changed something as a result. |
| Total | 100 |
Physicians who build things.
Every mentor is a physician with experience in health technology. They understand how healthcare works, which means you can skip the background and get straight to the part you're stuck on.
They circulate the room for the whole 9:00–11:30 work session. Two checkpoints are suggested on the schedule, but flag anyone down at any time — that is the entire reason they're here.
Open with your problem statement and one specific question. "Would you use this?" gets you a collegial yes. "What would make you close this and go back to what you were doing?" gets you something you can act on before 11:00.
That's a wrap.
Four hours, two tracks, and a room full of family physicians building what primary care needs next. Here's the morning in under a minute.
Three awards.
One winner in each track, scored only against the other teams in that track — plus one award that ignores tracks entirely and goes to whatever made us sit up. Announced Thursday, October 22, 3:30 – 5:00 PM at the Main Stage Event in the Karl Dean Ballroom.
Patient Empowerment Winner
Highest total rubric score among Track 1 teams. The build that most convincingly gives a patient more knowledge, agency, or control.
- Up to 5 free FMX 2027 registrations
Clinical Team Empowerment Winner
Highest total rubric score among Track 2 teams. The build that most convincingly gives a physician or care team their day back.
- Up to 5 free FMX 2027 registrations
The "OMG This Is So Cool" Award
For the build that made the room stop scrolling. Not necessarily the most rigorous or the most deployable — the one that made us stop and ask "wait, how?"
How the track winners are picked
Straight from the rubric. Every submission is scored across the five criteria, and the highest total in each track wins. There's no finalist round — every team is assessed the same way, from the same three minutes.
See the rubric →How the "OMG" award is picked
Deliberately not from the rubric. Once scoring is done, the organizers watch back the submissions and pick the one that genuinely surprised them. Ambition, audacity, and technical nerve all count.
Meet the sponsors.
The FMX Hackathon is an AAFP PCIN event made possible by organizations committed to advancing innovation in primary care.
In alignment with AAFP's principles, the Hackathon is intentionally designed to maintain a neutral, clinician-centered environment. Sponsors do not promote products or services during the event.
Questions, answered.
Everything we get asked before the doors open — most of it from participants who have never built anything before. If your question isn't here, ask an organizer at check-in or flag down a mentor.
Building without a tech background
5I'm not an engineer. Can I actually build something in one morning?
Yes — and this event is designed on that assumption. Nobody in the room is a professional software developer, so nobody is starting at a disadvantage.
What makes this possible now is that AI build tools have collapsed the distance between describing a thing and having a working version of it. People who have never written a line of code routinely produce a functioning prototype in an afternoon. Two and a half hours is tight but real.
And the thing you bring that can't be taught in a morning — knowing which problem is actually worth solving — is worth 40 of the 100 rubric points.
Do I need to know how to code?
No. Teams win on the clarity of the problem and the sense of the solution, not on implementation. Perfectly valid ways to build:
- AI build tools that generate a working app from a description
- No-code and low-code builders for forms, workflows, and simple apps
- Clickable design prototypes — Figma or similar, where nothing is real but every click works
- A narrated slide walkthrough of the interaction, if that's genuinely the best way to show it
The last one scores lowest on Prototype & video execution, so use it as a fallback rather than a plan. Mentors are here specifically to get you past whichever part you haven't done before.
What should I set up before October 20?
Half an hour the night before is worth an hour on the day:
- Create a free Loom or Tella account and record a ten-second test video. Confirm you can open the link in a private browsing window.
- Make sure your laptop can screen-share and that your microphone works.
- If you already use an AI build tool, log in and confirm you have credits.
- Think about one problem. Not a solution — a problem.
Does it have to be software?
No. A workflow redesign, a communication protocol, a new form, a service change — all valid, as long as your video shows it working in some form. A role-played patient conversation or an annotated workflow with a live walkthrough can clear the bar. What doesn't clear it is describing what you would have built.
What if we don't finish?
Submit anyway. A sharply framed problem with a half-working prototype and an honest account of what broke beats several polished entries that never named a real patient. Nearly every team feels unfinished at 11:30 — that's the format, not a failure.
Teams
4How do we form teams?
Team formation runs 8:30 – 9:00, immediately after the keynote.
It's an icebreaker activity — you'll meet people, find teammates, and start ideating on problems together. You don't need to arrive with an idea; most teams land on theirs during that half hour.
How big can a team be?
2–5 participants per team. Big enough to split the work across a two-and-a-half-hour build, small enough that nobody spends the morning waiting on a decision.
Do I need to know which track I'm in before I arrive?
No. Choose during problem selection and declare it on the submission form. If your idea spans both, pick the track whose primary user your solution serves. You can switch until you submit, but do it early — there's a winner in each track, and switching at 11:45 means you framed for the wrong audience all morning.
Can one team submit to both tracks?
No — one submission, one track, per team. Splitting your attention across two briefs in two and a half hours is the fastest way to place in neither.
Projects & eligibility
4Do solutions have to be built during the hackathon, or can I showcase an existing project?
They have to be built during the hackathon. Existing projects are not eligible for judging — this is a build event, not a demo day, and showcasing prior work isn't fair to teams starting from nothing at 9:00.
What you can bring:
- Generic scaffolding — starter templates, design systems, an AI tool you already pay for.
- Ideas and research. Thinking hard about a problem in advance is encouraged. Show up with a strong problem statement.
What you can't bring: a working prototype of your solution, problem-specific work done before the event, or a product you're already using in practice. The idea and everything that demonstrates it must be built between 8:00 and 12:00 on October 20.
Can I keep working on something I started at a previous event?
Not as a competing entry. You're welcome to attend and keep building, and mentors will happily give you feedback — but a continuation of prior work can't be scored against teams that started this morning. The better move is to take what you learned and attack a different slice of the same problem from scratch.
What if my idea is something a vendor already sells?
Then you've found a real problem, which is most of the battle. Say so in your video and explain why the existing product doesn't work in your setting — too expensive, wrong workflow, not available on your EHR, ignored by your team. "The thing exists but nobody can actually use it" is a legitimate and often excellent problem statement.
How finished does it need to be?
Much less finished than you think, and much more working than you think. One narrow interaction running end to end beats five features that are all 60% done.
Mocking parts is completely acceptable — hardcoded data, a fake API response, a human behind the curtain. Just say which parts are real in your video. That candour is rewarded; overclaiming is not.
Submitting your video
6Are there live demos?
No. Every team submits a recorded three-minute video between 11:30 and 12:00.
Submissions are reviewed after the morning ends. Winners are announced Thursday, October 22, 3:30 – 5:00 PM at the Main Stage Event in the Karl Dean Ballroom.
What should the video actually look like?
Narrate over your prototype while you click through it, or narrate over a short slide deck. Screen recording with your voice on top — that's it. No editing, no music, no title cards.
What if we run over three minutes?
We may not count any content beyond three minutes as part of the evaluation, so lead with what matters most.
Do we all have to be on camera?
Not required — but please do. Part of what makes these videos worth watching is seeing the people behind the ideas, and both Loom and Tella put your face in a small corner bubble while your screen fills the rest, so it costs you nothing.
It doesn't affect your score either way. One narrator is usually cleaner than four, so if the whole team wants to appear, a quick round of introductions at the start works nicely.
Do I have to use Loom or Tella?
No — any tool that produces a shareable link works. We recommend those two because they're free, browser-based, and generate a link the moment you stop recording. Whatever you use, the requirement is the same: the link must open without signing in.
Test it in a private browsing window before you submit. A link that asks for a login scores zero.
What counts as a supplemental file?
Anything that helps convey what you built: a live link to the prototype, a repository, your slide deck, screenshots, workflow diagrams, or notes from someone you tested it on. All optional.
Supplemental files are reviewed but never a substitute for the video. Assume the three minutes are all that gets watched.
Logistics
4What should I bring?
- A laptop and charger — you can't build or record on a phone
- Headphones with a microphone, which make your video noticeably clearer in a room of 100 people
- Any AI tool accounts you'll want, set up the night before
- One problem you actually care about
Breakfast, coffee, wifi, power, and snacks are provided.
What if I arrive late?
Come anyway. But the 8:00 keynote and the 8:30 team formation block are where teams and problems get set, so arriving after 9:00 makes it hard to join a team and very hard to submit competitively.
Can I attend remotely?
This is an in-person event. The mentor access and the hallway conversations are most of the value, and a four-hour format doesn't leave room for hybrid coordination. Plan to be in the room from 8:00 to 12:00.
Do I have to come back for the Main Stage Event?
Winning doesn't depend on attending. But awards and closing remarks run Thursday, October 22, 3:30 – 5:00 PM in the Karl Dean Ballroom, the winning videos get shown, and it's where the conversations about actually building these things happen. Come back.
Judging
3How are submissions scored?
Every submission is scored by AI, with human oversight. Your submission is evaluated against the five rubric criteria — 20 points each, 100 total. The organizers then review the results before anything is announced.
It felt like the honest choice for an AI hackathon: we're asking you to build with these tools all morning, so we use them too.
Does a person look at our submission at all?
Yes. The organizers review the scored results before winners are confirmed, and can step in where something is off.
When do we find out?
Thursday, October 22, 3:30 – 5:00 PM, at the Main Stage Event in the Karl Dean Ballroom. Scoring and review happen between the end of the hackathon and then.