My perspective
Designing at the atomic level of care
There was a time when healthcare software asked clinicians to become administrators: to stop, click, remember, prove, reconcile, and serve the system before they could return to the person in front of them, if only for 30 seconds to say thanks and see them off.
The interface became a tax on attention, and a tax on outcomes.
My thoughts here are far from a criticism of the people who built those systems. Healthcare is complex because care is complex. The system has to manage risk, privacy, compliance, evidence, funding, continuity, and accountability. It has to hold histories that stretch across years, sometimes decades. It has to work across people, places, moments, disciplines, and thresholds of trust.
But somewhere along the way, the system became the process. The work of documenting care started to compete with the work of giving care. As John Culkin wrote, channelling Marshall McLuhan: we shape our tools, and thereafter our tools shape us. We designed our world, and our world now designs us.
A new material for design
For Heidi, this means working at the atomic level of healthcare. It means changing and reimagining the smallest units of work: a conversation becomes a record, a record becomes a structured object, and a clinical moment becomes something that can be understood, reused, checked, moved, and acted on. The patient becomes a human, the clinician equally so.
The raw material of care is being reshaped, which means the material of design can change too.
Most organisations are rushing into AI with more urgency than understanding, mistaking adoption for advantage and speed for progress. They're worried they're going to miss out and that first-mover advantage is real for them. Resoundingly, it is not.
AI is a moment for restraint and conviction. The organisations that will thrive are the organisations that look beyond applying AI to what they already know, and instead reimagine what they know with AI.
Industrial designers learned to work with steel, glass, plastic, foam, fabric, weight, tolerance, and manufacturing. Digital designers learned to work with screens, flows, platforms, data, states, latency, and interaction. AI-native healthcare asks us to learn a new set of materials: clinical language, patient context, evidence, uncertainty, consent, compliance, risk, trust, time, and attention.
It also asks us to understand the weight of a diagnosis, the fragility of a handover, and the invisible burden of being a patient inside a system that may know everything about you and still not see you clearly.
The material has changed, so the design discipline has to change with it.
Trust is the real product
Design at Heidi cannot be limited to arranging interfaces around AI. It has to shape how these new materials behave. Nor can it allow for the “urban sprawl” of an AI-accelerated way of working, where everyone is suddenly capable of building. When everyone is capable, the risk of fracture grows exponentially. The risk of an experience not feeling like it was made by one mind is enormous. That, in one fell swoop, will dissolve trust.
Beyond the method of developing our product, we also need to balance an inordinate number of tensions.
What should the clinician see? What should remain quiet? How should uncertainty be expressed? When should the product ask for confirmation? When should it get out of the way? How does evidence appear without overwhelming the consultation? How does the product preserve clinical agency while reducing cognitive load? How does compliance become something designed into the system rather than bolted on afterwards?
These are not only UX questions. They are trust questions. Each one we get wrong erodes trust. Each one we get right changes the material.
A clinician will not trust Heidi because the interface is polished and beautiful (though it will be). They will trust Heidi because it behaves consistently when the stakes are high. Because the language is precise, the hierarchy is calm, and the small moments feel considered.
Because the AI is clear about what it knows and what it does not know. Because the product supports judgment rather than pretending to replace it. Because the clinician feels more present with the patient, not more entangled with and suffocated by the machine.
Trust is built in the small moments. It is built in loading states, empty states, error states, edit states, handoff states, citations, explanations, defaults, confirmations, permissions, and quiet moments of restraint. It is built, for instance, in how the product behaves when the AI is uncertain about what it heard: whether it flags the gap honestly, asks for a quick confirmation, or quietly guesses and hopes. The honest version costs a moment; the dishonest version costs the relationship. Trust is built when the product resists the temptation to be clever and chooses instead to be clear.
This is where design becomes infrastructure.
At Heidi's stage, design as a system cannot simply be a component library. It has to become a way of carrying judgment across the company. It has to encode taste, interaction principles, content patterns, accessibility standards, clinical workflow behaviours, evidence conventions, and product values.
It has to help many people build as if one clear mind was behind the work. The product should feel like it was designed by one person, even as it is built by many. That does not mean control. It means coherence, a clear narrative, and conviction.
The best design cultures do not scale through permission. They scale through clarity. People know what good looks like. They know which decisions matter. They understand the principles behind the patterns. They have rituals for critique, review, and improvement. They can move quickly because they are not debating the basics every time.
This is what great product organisations understand.
Airbnb is useful to think about here, not because Heidi should copy Airbnb, but because Airbnb showed that design can be the unifying thread between product, brand, service, culture, trust, and company strategy.
Airbnb did not only design a booking flow. It designed a new trust system between strangers. The product, language, photography, host tools, reviews, identity, service model, and brand all worked together to make an unfamiliar behaviour feel possible.
Heidi's challenge is different and more delicate. Airbnb helped strangers trust each other in unfamiliar homes. Heidi has to help clinicians trust AI with the context of care.
That requires design to operate at a different level. The interface is only the visible part. Beneath it sits clinical meaning, patient history, organisational risk, regulatory confidence, and the emotional reality of care.
The person inside the record
As I mention in my cover letter, I know that reality from both sides of my life.
Professionally, I have spent years helping teams make complex systems legible. Personally, I have lived my life inside healthcare. I know what it feels like when the system sees you as a patient record before it sees you as a person. I know the strange experience of being translated into notes, diagnoses, tests, risk factors, histories, handovers, and assumptions.
I also now know the relief of being seen properly.
That is the design standard I would hold. Heidi should not make clinicians more dependent on technology. It should make them more available to the human being in front of them. It should not turn patient data into a colder abstraction. It should help clinicians see the person inside the record more clearly. It should reduce the clinician's anxiety and set them free to do the job they so desperately want to do: care for others.
The opportunity at Heidi
A company building at Heidi's pace needs more than components, rituals, and roadmaps. It needs a strategic narrative strong enough to align the vision of its founders, product leaders, engineers, designers, clinicians, and customers around what the product is becoming.
What does it mean to be an AI Care Partner? What should Heidi never do? Where should the product be opinionated? Where should it preserve agency? How does brand show up inside workflow? How does compliance become a creative constraint? How does the product feel calm without feeling passive, intelligent without feeling intrusive, powerful without feeling unsafe?
More than design questions, these are leadership questions.
The best design leaders translate ambition into a form teams can act on. They turn strategy into principles, principles into patterns, patterns into decisions, and decisions into products people can feel.
With all this said and done, this is the opportunity I see at Heidi.
Not to make the product look designed, but to make care make sense again. To make trust scale. To turn new materials into humane software. To build a design culture that is fast, exacting, clear, and deeply sensitive to the stakes of the work.
Heidi is inventing a new way for clinicians to work with patient data, AI, evidence, compliance, and time.
I couldn’t think of a better design challenge to be involved in.