The numbers Brief Episode map Takeaways Charts Quotes Clips Glossary Go next Transcript

Episode 61 · Enterprise · 40 min

The constituency that cannot churn

Suki's bet is that healthcare's most powerful user is not the patient but the doctor — the one constituency that never leaves, generates almost all the data, and is burning out on what Soni calls unbelievably 90s-style software. Twenty percent of US GDP runs through healthcare and no Google-sized company has been built inside it; Suki's way in is a voice assistant that hands a clinician one to two hours of the day back.

Healthcare Product strategy AI & machine learning

Punit Soni

Founder & CEO, Suki.ai · with Vishal Krishna

39:32

Said in this episode

▶ 8:09

20% of GDP

Healthcare's share of the US economy

Bigger than e-commerce, search, social networking or media, Soni argues — and India's healthcare sector is 5-7% of GDP — yet neither has produced a technology company of Google's scale.

▶ 24:03

50-70%

Administrative time Suki claims to remove

Doctors using the product get roughly one to two hours of their day back, which they spend either on more patients or on going home — the retention argument.

▶ 24:32

$750,000

Cost to replace one departing US doctor

Attributed on air to a typical hospital study; Soni pairs it with the observation that doctors are currently leaving in groups.

▶ 24:47

~24%

Revenue lift from fewer insurance denials

Better-quality notes mean insurers deny fewer claims for the same patients; a separate ~19% rise in doctor wellness scores was also cited.

▶ 13:13

$10-15B

Market size per clinical skill

Soni's estimate for each of documentation, coding, order entry and clinical decision support individually — the roadmap after the note is captured.

▶ 31:58

~$100M

Raised in roughly five years

Soni declines to treat it as a milestone: a raise, he says, only means the company is not yet profitable enough to stop raising.

The brief

The argument in sixty seconds

Soni's claim is that healthcare has no Google-sized company because almost everyone attacks the wrong constituency. Twenty percent of US GDP — and five to seven percent of India's — runs through a sector bigger than search, e-commerce, social or media, yet patients care about healthcare only when they are sick, which makes them a user base engineered to churn. Doctors do not: they are always present, they create most of the data, and they are burning out on 90s-era software thickened by regulation and liability. Set that against a technology trend — voice today is where mobile was in 2005-06 — and you have the classic startup equation he borrows from the cloud-CRM playbook that built Salesforce. So Suki starts as a Siri for doctors that writes the note, ships features every other week the way a consumer app does, and then runs two arcs at once: more skills (coding, order entry, clinical decision support, each a $10-15 billion market of its own) and less dictation, ending in ambient capture where the clinician simply says 'Suki, pay attention'. The economics he quotes: 50-70% of administrative time removed, one to two hours a day back, a 24% revenue lift from fewer insurance denials, set against the $750,000 a US hospital spends replacing a doctor who quits. The strategic tell is whom he refuses to serve — he walked away from contracts that would have got Suki in the door by building for CIOs, and by year three doctors were telling their CEOs they would resign if Suki was not brought in. India is the larger bet: on Aadhaar-linked records and open APIs, he thinks Suki could be more than an assistant.

Worth your time if you are

Founders deciding which side of a market to actually serve

Product leaders putting voice or ambient AI inside a workflow

Health-tech operators tired of marketplace economics

Anyone building on India's Aadhaar-linked health stack

First-time founders about to celebrate a nine-figure round

Episode map

Where the conversation travels

Every block is a chapter, coloured by what it's about. Click any of it to jump straight to that minute on YouTube.

HealthcareProduct strategyAI & machine learningFounder journeySales, GTM & growthRegulation & policy

01Cold open: all of us are the same template 0:00 Vishal frames the administrative burden his surgeon wife carries before Soni waves off his own CV — NIT Kurukshetra, Wyoming, Wharton — arguing that middle-class Indian children are built from one template and the only real variable is how much risk you are willing to take. 02Google, Motorola, and the two-year bet 3:34 Product thinking came from Google's democratised, 20%-time culture and from a habit of volunteering for the unfashionable job — mobile before it mattered, then social, then a dying all-American Motorola that produced the Moto G, then Flipkart — every choice made against a guess about where the world would be in two or three years. 03Twenty percent of GDP and no Google 7:06 Healthcare is 20% of US GDP and 5-7% of India's — bigger than e-commerce, search, social and media — yet has produced no technology company of that stature, and its very difficulty makes time and resources a founder's advantage rather than a tax. 04Doctors, not patients: the constituency equation 9:54 The startup equation is a powerful constituency with a huge problem whose solution rides a technology trend — Salesforce's burnt-out CRM managers plus cloud — and in healthcare that constituency is doctors, not patients who churn the moment they get well and not the marketplaces Soni dismisses as glorified e-commerce shipping medicines like frying pans. 05A Siri for doctors, shipped every other week 14:14 Suki Assistant began as the smallest believable product tested on doctor friends pretending to treat him, then borrowed consumer cadence — features every other week, unheard of in healthcare — and the speech and command-intent engine underneath became Suki Speech Service, a platform other health-tech companies can embed. 06Two arcs: more skills, less dictation 18:00 One arc adds skills — documentation, coding, order entry, clinical decision support, each a $10-15 billion market — while the other moves from full dictation to command-and-control to flexible phrasing to ambient capture, illustrated by a surgeon's consent note that Suki learns to write from the conversation itself. 07The user is the clinician, not the CIO 20:48 Soni will listen to chief medical, financial and information officers but insists they are not Suki's user, a stance that cost him contracts in the early years and paid off by year three when doctors began telling their CEOs they would resign without it — and, he says, no pilot has yet been lost on the way to a commercial contract. 08Hours back, denials down, doctors staying 23:16 The US return-on-investment case: 50-70% of administrative time removed and one to two hours a day returned, a roughly 24% revenue increase as insurance denials fall on better-documented notes, about 19% higher wellness scores, and the $750,000 a hospital spends replacing a departing doctor. 09India's leapfrog and the open API 25:30 India moves from paper to something postmodern, skipping the generation the US is now stuck with, and Aadhaar-linked personal health records with push-pull APIs are the holy grail American healthcare will never crack — the reason Soni wants Suki to be a record layer in India, not merely an assistant. 10A voice in the room, not a favour 27:35 With a Bangalore team on the ground, Suki has begun early conversations in the health ministry — motivated, Soni says, less by advantage for Suki than by keeping India from repeating America's closed, non-interoperable systems, on the evidence that UPI proved this country can build open rails. 11A hundred million is not a milestone 30:45 Asked how he raised close to $100 million in five years, Soni refuses the compliment: money raised only means you have not yet built a company profitable enough to stop raising, and the real question is whether the capital has been converted into impact. 12Recruit smarter, then grind yourself down 32:45 Time management for a founder is really recruitment — hire people who know clinical workflows, machine learning and health-system selling better than you, keeping only the two or three problems nobody else can solve — while the emotional half means accepting that nobody cares about your pedigree and that persistence, not genius, is what he saw at Flipkart. 13Per doctor per month, and beachheads 35:18 Suki sells enterprise SaaS priced per doctor per month, will have to price India differently in pursuit of usage over margin, wants system-wide contracts covering doctors, nurses, PAs and MAs, and is chasing a few influential beachhead partners rather than broad coverage. 14Fifty in Bangalore; don't argue with fools 36:45 The India centre is about 50 people and aims to double by year end across AI/ML, mobile, backend, product and DevOps, before the episode closes on Soni's guide to happiness — pick your battles — and a reading diet kept deliberately far from business and technology.

Takeaways

Ideas to carry out of this hour

Solve for the constituency, not the disease

Soni's template for any startup, health or not: find a powerful constituency with a huge problem whose solution rides a live technology trend. Salesforce is the worked example — burnt-out CRM managers, cloud as the trend, and once the data flowed in, force.com and everything after. Applied to healthcare, the powerful constituency is not patients, who are interested only while sick and then effectively churn, and where the person who pays is a different, multi-variable party. It is doctors: permanently present, professionally demanding, and sitting on the data.

▶ 10:41 ▶ 10:56 ▶ 12:13

Most Indian health startups are e-commerce in a lab coat

Soni is blunt that many companies calling themselves healthcare in India are glorified e-commerce — sending medicines is the same operation as selling frying pans, and the infrastructure is a marketplace. A true healthcare company, on his definition, has to be hinged around clinical care itself, which means entering through the clinician's workflow rather than around it. The distinction matters commercially: marketplaces compete on logistics and discounts, while a clinical product compounds on data nobody else can generate.

▶ 11:43 ▶ 11:58 ▶ 12:13

Documentation is a wedge into $10-15 billion adjacencies

Healthcare data is created, more or less, by doctors — so a product that captures the note captures structured data at the source. From there Suki can attack coding (how insurance actually pays the doctor), order entry and clinical decision support, and Soni claims each of those phrases is a $10-15 billion market in its own right. The timing argument is that voice today sits roughly where mobile sat in 2005-06, which is why he thinks the wedge works now and did not five years earlier.

▶ 12:43 ▶ 13:13 ▶ 18:46

The product's second arc is subtraction, not features

Arc one adds skills; arc two removes the user's effort. Early Suki was dumb, so a surgeon had to dictate the whole risk-and-benefit consent every time; then it learned that surgeon's consent as a template invoked by command; then it accepted any phrasing of the same intent; and the last step is ambient — 'Suki, pay attention' — where the note is assembled from the conversation with the patient. Soni will not promise what share goes ambient, but argues that even 10% ambient is 10% of the doctor's work simply gone, with command-and-control covering the rest.

▶ 19:16 ▶ 20:02 ▶ 20:33

Know your user well enough to refuse the deal

Soni's rule is that every strategic decision gets tested against one question: does this benefit the doctor? That meant listening politely to chief medical, financial and information officers while telling them they are not the user, and walking away from early contracts that demanded features which would have opened the door but not served clinicians. The payoff arrived by year three as a groundswell — doctors telling their health-system CEOs they would resign if Suki were not brought in — and he claims no pilot has since been lost on the way to a commercial contract.

▶ 21:47 ▶ 22:18 ▶ 22:47

India is a leapfrog bet, not a discounted US market

Soni expects India to jump from paper straight to something postmodern, skipping the generation of systems the US now cannot delete. Aadhaar-linked personal health records with APIs you can push into and pull from are, in his words, the holy grail American healthcare will never crack — and because a voice product generates structured data from the horse's mouth, Suki wants to feed those rails. Hence the ambition gap he states plainly: in the US Suki will be a digital assistant, in India it has a shot at being a new kind of medical record system on top of the national APIs.

▶ 25:47 ▶ 26:02 ▶ 26:35

Money raised is an admission, not an achievement

Close to $100 million in five years earns no self-congratulation from Soni: in this ecosystem people raise a lot of money, sometimes with little to show, and a raise only means you have not yet built a company profitable enough to stop raising. He calls it a necessary evil and redirects the question to whether the capital produced impact. The founder's actual leverage, he argues, is recruitment — hire people who beat you on clinical workflows, machine learning and health-system selling, and you buy back time for the two or three problems only you can solve.

▶ 31:58 ▶ 32:14 ▶ 32:46

The numbers, drawn

What the episode measures

Every figure below was said on air — timestamps included, caveats kept.

Conversation share

portion of the hour spent on each theme

Healthcare19%

Product strategy16%

AI & machine learning13%

Founder journey13%

Sales, GTM & growth10%

Regulation & policy9%

Computed from the chapter map of this episode.

What Suki says a health system gets back

% improvement claimed

Vendor-stated figures from the conversation: 50-70% of administrative time saved (lower bound plotted), ~24% higher overall revenue for the same patients as insurance denials fall, ~19% higher satisfaction or wellness levels. Percentages measure different bases.▶ 24:03

Healthcare's share of GDP, as stated on air

% of GDP

Soni's numbers in conversation: healthcare is 20% of US GDP and 5-7% of India's — the India bar plots the midpoint of that range.▶ 8:09

Worth keeping

Lines that stay

The most powerful constituency in healthcare is not patients. Patients as a consumer class are only interested in healthcare when they're sick — and when they're not sick, it's over.

— Punit Soni ▶ 11:11

With all due respect to chief medical officers, chief financial officers, chief information officers, CEOs of health systems — we will listen to you, but you're not our user. Our user is the clinician.

— Punit Soni ▶ 21:33

By year three, doctors were telling their CEOs they would resign and leave the health system if Suki was not brought in.

— Punit Soni ▶ 22:33

I don't think raising money is anything to feel good about. Money raised only means you haven't yet built a company profitable enough that you don't need to raise money.

— Punit Soni ▶ 31:58

Nobody cares how great you are or what pedigree you have. You have to grind yourself down and become the simplest version of yourself — and then maybe you have a shot.

— Punit Soni ▶ 33:45

Clips that travel

Short on time? Start here

Founders deciding which side of a market to serve

Why doctors and not patients

The episode's thesis in one stretch: the constituency equation, the Salesforce parallel, and why patient-facing health startups are marketplaces in disguise.

9:54 → 14:14 · 4 min ▶ Watch clip

Product leaders building AI into a professional workflow

From dictation to 'Suki, pay attention'

The full build story — smallest believable product, consumer release cadence inside healthcare, the speech platform, and the four-step march to ambient capture.

14:14 → 20:48 · 7 min ▶ Watch clip

Enterprise sellers stuck in health-system procurement

The user is the clinician, not the CIO

Refusing revenue to stay loyal to one user, and the year-three moment when doctors threatened to resign over a piece of software.

20:48 → 23:16 · 2 min ▶ Watch clip

Anyone building on India's national health stack

The India leapfrog

Aadhaar-linked records and open APIs as the holy grail the US will never crack, and why Suki wants a voice in the ministry before the rails set.

25:30 → 30:45 · 5 min ▶ Watch clip

First-time founders raising a large round

A hundred million is not a milestone

The least celebratory fundraising answer you will hear, plus time management as recruitment and the case for persistence over genius.

30:45 → 35:18 · 5 min ▶ Watch clip

Glossary

The jargon, unpacked

EHR / EMR

The electronic health or medical record — the system of record a doctor must type into after every encounter, and the software Soni describes as 90s-style and unwieldy.

Ambient documentation

Software that assembles the clinical note by listening to the doctor-patient conversation itself, rather than waiting for dictation or commands — the endpoint of Suki's second product arc.

Medical coding

Translating an encounter into standardised billing codes; it is how insurance decides what the doctor gets paid, and one of the adjacent skills Suki plans to automate.

Claim denial

An insurer's refusal to pay for a documented service, often on paperwork grounds — Suki's revenue claim rests on better notes producing fewer of them.

PHI

Protected health information — patient data whose handling is legally restricted, and the compliance layer Suki had to build before its first clinic pilot.

Ayushman Bharat / Aadhaar-linked health records

India's national health programme and the digital record layer Soni describes, where a citizen's medical record is tied to Aadhaar and reachable through APIs that let systems push and pull data.

Health-data interoperability standard

The common format that lets different hospital systems exchange records; Soni argues India's APIs must adopt the standard everyone else uses, though the auto-captions garble its name.

Beachhead

A small set of large, influential early customers chosen to learn from rather than to maximise revenue — Suki's stated India strategy.

Connections

If this resonated, go here next

EP 46 · Deep tech

Full transcript

The whole conversation, searchable

Auto-generated captions, lightly cleaned. Click a timestamp to open that moment on YouTube.