Insights
How do I bring AI into my practice without disrupting patient care?
Start where your patients aren't. The safest first steps for AI in a healthcare practice are marketing, team training, and paperwork — the work that surrounds care, not the care itself. Match each tool to one real pain point instead of buying a one-size-fits-all system, and treat prompting as a learnable clinical skill. Do that, and nothing about the patient's experience is interrupted — except that their care team is more fully present with them.
Where should AI start in a healthcare practice?
Not in the treatment room. The treatment room is where trust lives, and trust is the last thing you experiment with.
Every practice carries a second, quieter workload: attracting patients, onboarding and training the team, and the paperwork that follows every visit like a shadow. National data says that shadow is long — practitioners now spend roughly two hours on records and desk work for every hour of direct patient care. That second workload is where AI belongs first: real value, zero risk to the moment of care.
From the field
We recently began working with a large group practice. The worry in the room was never a robot replacing anyone. It was quieter and more honest: "How do we even start? How do we ask this thing well?" So we didn't start in the clinic. We started with marketing, team culture and training, and paperwork — and we made prompting itself the first skill we taught. The lesson we keep relearning: the barrier to AI is almost never the technology. It's the first conversation with it.
Why doesn't one-size-fits-all AI work?
Because no two practices hurt in the same place. One drowns in referral letters; another loses new patients to unanswered phones; a third has a training binder nobody opens. When AI is understood well, it can be aimed at your unique pain points — precisely, one at a time — rather than draped over the whole practice like a tarp.
The profession is already moving this way. In the AMA's latest national survey, 81% of physicians reported using AI in their work — up from 38% in 2023 — and the average user runs more than two distinct use cases. Adoption is not one big system; it's several small, well-aimed ones.
Aimed well, every recovered hour flows back to the same place: the patient in front of you. Routine work costs less energy, so presence costs less effort.
15–20%
of weekly staff hours recaptured from routine work in our client engagements — time that returns to patient care. Your number depends on your starting point; we'll tell you honestly what to expect.
What's the business case for lifting the routine workload?
There are two ledgers here — the economic one and the human one — and they turn out to be the same ledger.
The economic side is stark. Burnout driven largely by administrative load costs U.S. healthcare about $4.6 billion a year — roughly $7,600 per employed physician, every year — through turnover and reduced clinical hours. Every routine task you lift off your team is a quiet payment against that bill.
The human side pays even better. Patients can feel the difference between a practitioner who is halfway into a screen and one who is fully present — and that presence, that heart connection, is what they remember, tell their friends about, and return to. It isn't soft sentiment; it's the strongest business asset a practice owns. Deloitte's analysis of patient-experience data found hospitals with excellent patient ratings earned net margins of 4.7%, versus 1.8% for those rated low. More recent research confirms the direction: better patient-reported experience is associated with higher future revenue and lower costs.
Presence heals, and presence compounds. The routine workload is what erodes it — one interrupted glance at a screen at a time. Lift the routine, and you aren't just saving hours; you're buying back the very thing patients came for.
How do I train my team without slowing care down?
Treat AI like a new team member: onboard it away from the treatment room, introduce it to patients later. Three steps carry most of the weight:
- Map the pain points. List the ten tasks your team dreads. Circle the ones that never touch a patient. That circle is your starting lineup.
- Pilot one workflow. One tool, one task, one owner, two weeks. Marketing copy, a training outline, a stack of routine forms. Small enough that failure costs nothing; real enough that success is felt.
- Train prompting like a clinical skill. Here is the honest part: talking to AI is not just talking to another person. Prompting is a science — closer to writing a precise referral than to casual conversation. Context, constraints, and examples change everything. The good news: like any clinical skill, it is absolutely learnable, and most teams get genuinely good in weeks, not years.
Notice what never appears in those steps: any pause, any interruption, any experiment in front of a patient. Care continues untouched while capability grows around it.
Why start now?
We believe this is a unique inflection point in human history. For the first time, this capability is not reserved for a few institutions with research budgets — it is reachable, learnable, and affordable for every healthcare practice. Everyone has the capacity to build their dreams right now. It is no longer for just a few in the world. It is for all.
Start away from the treatment room. Aim at one pain point. Learn to ask well. The rest follows — and what follows is more presence, more heart, more of the medicine only humans can give.