Bringing Unseen Patients Into the Conversation
If you’ve ever tried to manage a complex patient problem through a legacy portal at 11:07 p.m., you know the vibe: messages disappear, attachments fail, and the “urgent” flag is mostly decorative. Drs. Arnold Lee and Benjamin Bergman built VocalMD because they’ve lived that chaos—from One Medical to ER trenches—and decided the status quo is not just clunky; it’s unsafe and inequitable.
Here’s the blunt version of their talk, tailored for founders and operators who ship.
The Problem (it’s not subtle)
Fragmented care + dead-end messaging. Patients bounce between systems; clinicians can’t see or coordinate in real time. Email gets spam-filtered; portals aren’t conversational.
High-need patients fall through the cracks. Unhoused patients and those with substance use disorders ping-pong through ERs with catastrophic cost and terrible outcomes.
Perverse incentives. Systems optimize for visits and billing, not continuity and outcomes. Prior auth and network carve-outs add friction exactly where you need speed.
“I had patients I knew I could help—if I could actually talk with them when it mattered.” —Dr. Lee
The VocalMD Thesis
Make ongoing, two-way, team-based communication the default—not the exception.
VocalMD is a mobile-first layer that lets clinicians and patients message, refer, and coordinate in real time, then pull records via FHIR where possible, and retain continuity even when care sites change.
Key building blocks:
Clinician network access: Single sign-on through Doximity enables millions of verified clinicians to log in and participate.
Double opt-in panels: Patients choose a physician; physicians accept—then one-click referrals add specialists, mental health, nursing, and social work into the same thread.
FHIR ingestion (pragmatic): Pull what you can, when you can. It’s imperfect in 2025, but every usable datapoint helps the care thread.
Asynchronous + live care: Text when it’s simple; jump to a tele-visit when it’s not.
What it’s not: Yet another portal bolted to an RCM engine. The care relationship—not the billing code—is the product.
Who Benefits First
While VocalMD is for everyone, the first beachhead is the population with the most to gain and the least support: unhoused patients and people with SUD. The model: meet them where they are (ERs, shelters, street-medicine teams), get them logged in once, and keep the thread alive across devices, clinics, and crises.
Pilot signal: In an early nonprofit cohort (10 patients, rural, car-living unhoused), they saw one ER admission in a year—anecdata, yes, but directionally compelling.
Regulatory Reality (and how to navigate it)
HIPAA & privacy: Individual clinicians remain covered entities; the app must be secure and disciplined. If a patient consents and records are handled properly, communication is permissible.
FHIR & information blocking: Enforcement is uneven. Expect friction (including dramatic on-screen “warnings”). Design workflows that work even when FHIR doesn’t.
Prior auth & burnout: Don’t romanticize AI away. The founders are focused on human trust with automation behind the scenes, not chatbots practicing medicine.
Business Model (because mission needs a margin)
Cash membership for private patients (priced to be accessible).
Traditional tele-billing where appropriate (PC model).
Value-based / fixed-fee pilots with payers and systems to reduce readmits and ER churn.
For the unhoused: the app fee is donated; funding flows via grants, payers, and health-system partnerships.
AI: Useful, Not Worshipped
VocalMD is experimenting with agentic AI for data capture, analytics, and workflow automation—to measure outcomes, prove cost savings, and scale the back office. They’re not handing clinical judgment to a hallucination engine. Smart.
GTM: Where They’re Aiming
ER intake → instant enrollment. Turn “here’s a sandwich” into “here’s your care team.”
Street medicine kits. A backpack, an iPad, and one-tap specialist consults on the curb.
Shelter & nonprofit partners. The most credible stakeholders for unreachable patients.
Municipal & payer pilots. Follow the money: reduced readmits, fewer avoidable ER visits, MIPS improvements.
Hard truth: Big systems protect their networks and status quo tooling. Evidence + cost math will convert them faster than morality plays.
What MedStart Founders Can Steal
Design for the 11:00 p.m. moment. Real care is asynchronous, messy, and mobile.
Make referral a click, not a committee. Network effects require zero-friction invites.
Collect policy-grade data from Day 1. If you can’t prove dollars saved and outcomes improved, you’re a nice demo.
Respect the human layer. AI can accelerate ops; it cannot replace trust with traumatized patients or burned-out clinicians.
The Ask
Intros: ER groups, shelters, street-medicine teams, and health plans willing to pilot.
Pilot sites: Sacramento, SF Bay Area, Northern NJ.
Capital: Targeting $3M to scale pilots, analytics, and nonprofit deployments.
If your organization wants better outcomes and fewer midnight spirals through the portal maze, this is the kind of product to test—where care continuity is the feature, not the afterthought.
