Case study 2 of 5 · Decided by approval
A physician calling a patient from a cell phone could give the number away for good, or block it and show “Unknown Caller,” which patients are less likely to answer.
The patient sees the office, not the cell phone
As founding product lead, I originated Doximity Dialer’s identity layer: a physician could call from their own phone while the patient saw the office number.
- My part
-
Founding product lead, Doximity Dialer
- Originated the identity layer that lets a physician call from their own phone while the patient sees the office number.
- Led the HIPAA compliance policy and the EHR integration strategy.
- The call
- Shipped caller ID through a vendor that didn’t require proof of number control, with Doximity’s account-level identity verification as the protection underneath.
- Turned down
- Twilio-verified caller ID. Verified, but slower to ship.
- Result
- More than 300,000 calls on an average workday across 250+ hospitals and health systems, 9 years on.
The patient sees the office, not the cell phone · Kanav Jain · kanav.net
-
300,000+
Calls on an average workday
Doximity’s own figure, February 2026.
-
250+
Hospitals and health systems
From the same Doximity release.
-
9 years
Still running since I left in 2017
Originated as founding product lead; the caller-ID route was unchanged when I left.
Context
Physicians need to reach patients but face a bind: call from the hospital desk phone, which means being at the hospital, or from a personal cell, which gives the number away for good, with no boundary, no audit trail, and a safety risk. Most choose neither, and the call doesn’t happen.
I did not believe we could get that through a hospital IT security review as a proposal. A physician placing outbound calls under an office number from a personal phone was something those reviews had no category for, and I expected to lose the argument on paper. Authorization did come after the product was in use, and in stages.
The decision
I proved the caller-ID route over a weekend and launched on a vendor that did not require proof of number control. Hospital authorization followed adoption; the route was unchanged when I left.
Twilio-verified caller ID Not chosen
Named, with pricing, in the April proposal as “our chief 3rd party vendor.” Every number had to be proven before it could be displayed; launch would have waited on verification.
Vendor without proof of control Chosen
Named in the August spec. The physician picked the number to show, the office line by default, and the verified account behind the call carried the protection.
The displayed number belonged to the practice or hospital. The protection was Doximity’s verified physician account: each call could be traced to its caller. This did not involve an automated clinical decision.
| Compared on | Hospital desk phone | Personal cell | Dialer |
|---|---|---|---|
| Where you must be | At the hospital | Anywhere | Anywhere, on your own device |
| What the patient sees | The hospital’s number | The personal number, given away for good, or “Unknown Caller” if it is blocked | The office number, and any name the patient saved for it |
| Audit trail | — | None | Audit-ready call logs and identity verification records |
| Whether the call is answered | — | Less likely to be answered; most physicians choose not to call at all | A number the patient knows |
What happened
I designed and shipped the HIPAA-compliant caller-ID and fallback workflows, so the clinician’s office was the number the patient saw by default.
I worked with legal and hospital IT to clear EHR integration across several systems, including the Epic Haiku integration for one-tap calling inside the chart. 5 6
The identity layer shipped, and Dialer scaled on it. The launch cohort converted: of 904 physicians emailed, 36 placed a call within days, and daily calls went from roughly five to about 680 in the first eleven days.
| Identity | What it is | Where the number comes from |
|---|---|---|
| Office | The clinician’s office number — the preselected default | An existing line the clinician already uses |
| Back office | A back-office line, selectable per call | An existing line the clinician already uses |
| Mobile | The clinician’s own mobile number, chosen deliberately | One the clinician already holds |
| Ghost | A temporary number the product issues when none of the clinician’s existing numbers fits | Issued by the product; expires after 24 hours |
| Method | What it proves | What the spec says about retention |
|---|---|---|
| Medical email domain | Clinician status, by an address at a medical institution | Not stated either way |
| DEA number | Clinician status, by the prescriber registration | No — used to verify, never stored |
| License photo | Clinician status, by a photograph of the license | Not stated either way |
| Fax | Clinician status, by a faxed document | Not stated either way |
What I’d keep
Adoption may come before official approval only while the risk stays with the person choosing. When the risk is clinical, approval comes first.
The first sign it was working wasn’t a metric: medical residents I’d gone to undergrad with told me they were getting more sleep at night, since they could call patients from home with the office number instead of staying at the hospital to use the desk phone.
Next step
Advisory and fractional product leadership for healthcare and AI ventures. I partner with technical founders to take complex clinical systems through review, unblock patient adoption, and build products that last.
Sources
1 Dialer launch
Doximity Dialer launched in 2016 as a physician calling product.
Doximity’s launch announcement.
2 Launch rationale, in my words
I stated the product’s rationale on Doximity’s blog at launch: “Easier communication leads to more communication, and greater communication leads to better health outcomes.”
The original post is offline; this is the Internet Archive copy, dated 13 December 2016, and it quotes me as Product Manager, Growth and Mobile Apps. The title names what Dialer replaced: physicians blocking caller ID with *67, and patients not answering.
3 Dialer call volume
Doximity reported in February 2026 that Dialer carries more than 300,000 calls on an average workday, across 250+ hospitals and health systems.
Doximity’s own figure, from its investor newsroom.
4 Telehealth provider volume
Over 300,000 unique active providers used Doximity telehealth tools in the quarter ended March 31, 2021.
From Doximity’s SEC filing: “We had over 300,000 unique active providers use our telehealth tools in the quarter ended March 31, 2021.” It counts people, not calls, across all of Doximity’s telehealth tools in early 2021, and is not the workday call count above. Its count of providers includes physicians, doctors of osteopathy, physician assistants, nurse practitioners, and medical students.
5 Epic Haiku integration
Dialer integrated with Epic Haiku.
Doximity’s announcement of the integration.
6 Epic Showroom listing
Dialer appears in Epic Showroom.
Dialer’s listing in Epic’s app marketplace.
Documents described, not republished
- Caller ID mapping The August 2016 MVP spec, when the product was still called CallPatient (tabulated above).
- The decision to defer phone verification In the August 2016 MVP spec, four SMS-verification screens are struck through in red with the contemporaneous note: “Let’s defer verifying the user’s phone until version 1.1… We do already have them in as a registered & verified doximity user so that’s a pretty good protection layer.” The account already carried the verification; reverifying was friction without protection.
- Identity verification methods The annotated iOS spec, tabulated above.
- HIPAA policies, procedures, and white paper Written in 2014 for the messaging products.
Launch figures and the weekend prototype come from my launch review. The 2026 call volume is Doximity’s own reported figure and is evidence about the product, not about which of its parts are still mine: I can say the caller-ID route was unchanged when I left in 2017, and nothing public says what the identity layer looks like now.
Revised