Hey,
Everyone wants to know: "What should we use ChatGPT for?"
Wrong question.
Here's the right one:
"What repetitive decision-making work is killing our team's time?"
Let me show you the difference.
The Problem Everyone Gets Wrong
Last month, a concierge practice owner asked me: "How do I use AI to write better job descriptions?"
I said, "You don't need better job descriptions. You need to stop wasting 25 hours interviewing the wrong people."
Same tech. Completely different outcome.
How We Actually Use AI at PURE
When we were hiring our concierge RN, I didn't start with the technology. I started with the problem:
The Problem:
185 candidates applied
Most looked good on paper
90% were terrible cultural fits
We were drowning in "highly qualified" people who'd never work
What everyone does:
Use AI to write a better Indeed post, then manually screen everyone anyway.
What we did:
Built an AI that filters by cultural fit BEFORE we ever talk to them.
We taught it to score every candidate on:
Cultural fit (40%)
Technical skills (35%)
Growth potential (25%)
Only candidates scoring 4.0+ moved forward.
The result?
Instead of interviewing 20 mediocre candidates, we interviewed 6 exceptional ones.
The AI didn't replace hiring. It replaced the part of hiring that was wasting our time.
The Framework
Here's how to actually implement AI in your practice (or business, or life):
Step 1: Find the Low-Value, High-Volume Work
Not "What's hard?" — What's repetitive and takes forever?
Examples from healthcare:
Screening resumes
Summarizing patient charts
Answering the same insurance questions
Scheduling follow-ups
Writing referral letters
Ask: "What work drains time but doesn't require human judgment?"
Step 2: Define Your Criteria
How do YOU make that decision?
When you're screening a resume, what are you actually looking for? Write it down.
For us, it was:
Do they use words like "empower" and "optimize"? (cultural fit signal)
Do they have concierge or boutique medical experience? (technical skill)
Are they early-career enough to grow with us? (growth potential)
Most people skip this step. They hand AI a vague task and wonder why it fails.
Define the rubric a human would use. Then let AI apply it at scale.
Step 3: Let AI Do the Grind
Now — and only now — you automate.
We didn't ask AI to "find good candidates."
We asked it to:
Read every resume
Score it against our rubric
Flag anything above 4.0
AI is incredible at applying consistent criteria to high-volume work. It never gets tired. It never skips someone because it's 4:45 PM on a Friday.
Step 4: Humans Do the Judgment Calls
AI gets us from 185 candidates to 6.
We pick the final hire.
That's the part that actually requires human intuition — the energy in the room, the way they answer a curveball question, whether we'd trust them with our patients.
AI amplifies judgment. It doesn't replace it.
What This Looks Like in Practice
Since we implemented this:
Hiring:
Time per position: 25 hours → 6 hours
Quality of final candidates: Dramatically higher
Decision fatigue: Gone
Patient intake:
AI summarizes new patient forms before the appointment
Dr. Miranda walks in already knowing the story
Patients feel heard faster
Lab analysis:
AI flags trends in bloodwork (e.g., "Ferritin dropping for 6 months")
We catch things earlier
Longevity members get proactive interventions, not reactive ones
None of this is "AI replaces the doctor."
It's "AI handles the grind so the doctor can be a doctor."
The Big Miss
Most healthcare AI projects fail because they start with:
"Let's buy this AI tool"
"Now what do we use it for?"
That's backwards.
Start with:
"What's killing our time?"
"What criteria do WE use to make that decision?"
"Can AI apply those criteria at scale?"
"Great — now humans focus on judgment calls"
Stop asking "What can AI do?"
Start asking "What shouldn't humans be doing?"
One More Thing
If you're building AI tools for healthcare and want real practitioner feedback, DM me. I'm happy to be a sounding board.
I've seen what works and what's BS. And I'll tell you the difference.
Also: I'll be writing more on this in Forbes soon. First piece drops next week on why most healthcare AI implementations fail (and how small practices like ours are quietly winning).
More soon,
Dustin
P.S. — That "unicorn RN" we hired? Emily scored a 5.5 on resume, 4.5 on the AI phone screen. She's MSN-FNP from U Miami, lives in Coral Gables, has concierge infusion experience.
The AI didn't find her. It just made sure we didn't miss her in a pile of 185 resumes.
That's the whole point.
The Mangas Brief
AI, medicine, and cutting through the noise.
dustinmangas.com