Musings on healthcare IT, FHIR, EHR systems, and digital health innovation

🔥 FHIR in the hole!
Back to posts

Share Your Health Record Like a Boarding Pass: AI Agents Help "Kill The Clipboard"

Original on LinkedIn

In my previous update on the Kill The Clipboard (KTC) initiative, I outlined our primary goal: replacing the repetitive, frustrating process of clinic intake with a seamless digital handoff. Patients should be able to share their health data—structured FHIR resources, clinical notes, and their own personal narrative—from an app they already use. The experience should be as simple as presenting a boarding pass at the airport.

Video
Watch on YouTube · Original on LinkedIn

However, a major practical challenge remains. How can we make it easy for patients to curate years of complex medical records into a concise, relevant package for a specific doctor’s appointment? Expecting patients to manually sort through raw data files or search for specific progress notes from years ago is not a realistic solution.

To address this, I have been developing a method that integrates AI agents into the workflow. Ahead of our panel session at DevDays in Minneapolis next week (with Anthony Pizzi, e-Patient Dave, and James Cummings), I have recorded a demonstration showing how we can use large language models, like Claude, to do this heavy lifting.

Here is a look at the new Kill The Clipboard AI Agent Skill.

The Demonstration: Preparing for a Neurology Consult

Using an AI environment, a patient can upload the new KTC skill.zip file (which contains instructions and backend scripts). This provides the AI with the specific instructions it needs to act as a privacy-preserving medical records assistant.

In the

, I first pulled 8 years of my own health data from a local provider using SMART on FHIR tools. Then, I gave the agent a simple instruction:

“Help me prepare for a neurology consult.”

Here is how the AI agent responded:

  1. Reviewed the Standards: The agent read the Kill The Clipboard specifications to understand exactly what a SMART Health Link (SHL) is and what data the clinic will expect to receive.
  2. Curated the Data: It scanned my 8-year clinical history and filtered out routine check-ups. Instead, it identified a concussion I had in 2020 and pulled the relevant spine x-rays, occupational therapy notes, and brain imaging.
  3. Drafted the “Patient Story”: A core requirement of the KTC specification is allowing patients to share their own narrative in their own words. The AI drafted a concise “Patient Story” from my perspective, summarizing the lingering effects of the concussion and my goals for the upcoming neurology visit.
  4. Generated the Secure Link: Finally, the agent packaged the structured FHIR data, the clinical progress notes, and the Patient Story into an encrypted bundle, generating a secure SMART Health Link.

The Provider Experience and Patient Controls

The output of this curation process is a simple QR code or URL.

When the patient arrives at the clinic, the front desk can scan the QR code. If the clinic’s Electronic Health Record (EHR) is fully integrated, the structured data (Problems, Allergies, Medications, Immunizations) and the Patient Story flow directly into the patient’s chart.

If the EHR does not natively support SMART Health Links yet, the link opens a secure web viewer built as a fallback. This allows the clinician to read the Patient Story, review a generated summary PDF, read the curated progress notes, and view the raw FHIR resources.

Crucially, the patient remains in control of their data at all times. The KTC skill provides a control panel for the generated link where the patient can:

  • View an Access Log: See exactly who opened the record (e.g., “Example Doctor”) and when.
  • Pause Sharing: Temporarily disable the link.
  • Revoke Access: Permanently disable the link and delete the hosted data when the visit is over.

Next Steps

By combining the standardized infrastructure of SMART Health Links with the reasoning capabilities of AI, we can significantly reduce the friction of health data sharing. The AI acts as a personal archivist, ensuring the clinician has exactly the context they need, perfectly formatted, before the visit begins.

Resources & Links:

If you are attending DevDays in Minneapolis next week, please join our panel session. We will be discussing the cultural and technological shifts required to make this workflow the new standard of care.

Conversation on LinkedIn

15 comments · 80 reactionsReply on LinkedIn
  • Principal Healthcare Architect | FHIR + Epic Integration | HL7/Kafka Interoperability | Domain-Driven Design (DDD) + Adapter-Pattern Architecture | Patient Identity | | AI-Aligned Clinical Workflows

    As always inspiring and insightful @Josh Mandel

    1 reaction

  • Paul SwiderJun 12, 2026

    Mission Critical Healthcare Solutions | Strategic Microsoft Cloud and AI Partner

    Like always, great work. I'm excited to dive into this later. Thank you for all you do!

    2 reactions

  • Eugene Vestel, MBAJun 12, 2026

    Building enterprise AI platforms & agentic systems | Founder, HealthClaw | Host, Out of the FHIR podcast | FHIR + agentic memory + governance | Healthcare data builder, 15+ yrs

    Josh Mandel, MD this is great I just deployed using healthclaw.io and used it with Sally my telegram PCP bot to create a QR code for my checkin.

    Comment image, no alternative text available

    1 reaction

  • Eyal D.Jun 13, 2026

    Data Innovation Leader | FHIR and Interoperability evangelist | Data research platforms expert | A champion for Data

    Loved the demo. Takes things from technical to practical level.

  • @Trifork

    You can also maybe not "kill the portals" but then ... enhance them ... with this ... https://github.com/c3po-initiative/add-ons 😄

    3 reactions

  • Brian AhierJun 13, 2026

    Data & Quality Manager at West Virginia Department of Health

    Josh Mandel, MD is one of the most prolific developers of patient-centered applications and has an amazing way of producing technology that not only meets policy objectives, but actually helps people in their everyday lives. Thanks Josh!

    6 reactions

  • Uri GorenJun 13, 2026

    Head of Innovation @Meir | Leading patient-centric innovation with AI and design thinking expertise.

    You are really doing amazing work! Inspiring

    1 reaction

  • Mary Jane BogackiJun 15, 2026 · edited

    Certified Coding Specialist (CCS)

    The thoroughness of instructions for much of the time consuming “stuff” is commendable! I dislike the hours spent configuring new downloads/installations.
    What audience are you targeting? i.e. Will this be a one-step, fully automated download? Or will it require subscription to and familiarity-with-using agents and models?

    • Josh Mandel, MDAuthorJun 15, 2026

      Anyone can try it -- but I generally look at this as a tool for people who already know something about AI agents and how to point them at skills. I would love feedback on how things go if you give it a shot.

      1 reaction

  • A JJun 24, 2026

    --

    Josh Mandel, MD In a short clinic visit, how do you process all the information from the QR code and determine what's relevant in the moment? Some of us prefer to review the chart and prepare beforehand—is there a way to access this information before the visit? Thanks.

  • Ryan RadeckiJun 26, 2026

    Emergency Medicine Specialist | Clinical Informatics Director | Health NZ HealthX

    ... it sounds as though you're describing what ought to be implemented by the International Patient Summary standard? https://build.fhir.org/ig/HL7/fhir-ips/en/

    • Dr Chris EdwardsJul 1, 2026

      CMIO | Emergency Physician | Board Director | Digital Health Clinician Advocate

      Ryan Radecki yay for IPS call out :)