AI Is Rewiring Pregnancy Care: Babyscripts CEO on RPM, Risk, and Real Outcomes
with Anish Sebastian, Babyscripts
AI Is Rewiring Pregnancy Care: Babyscripts CEO on RPM, Risk, and Real Outcomes
Show Notes
Maternal deaths in the United States are rising. The US is the only advanced economy where that is happening. Not in a country without hospitals or insurance - here, in one of the richest countries in the world, women are dying from complications that other nations have learned to prevent. And one of the core drivers is not clinical failure. It is access failure. Mothers are not being monitored closely enough between appointments, risks are escalating without detection, and a system still running on fax machines cannot see what is happening at home.
Anish Sebastian saw this problem in 2013 and built Babyscripts around a single conviction: the best time to catch a pregnancy complication is before it becomes an emergency, and that requires data collected outside the clinic. Over the past twelve years, Babyscripts has raised more than $37 million, managed hundreds of thousands of pregnancies, and built a remote patient monitoring platform that ships connected blood pressure cuffs and cellular devices directly to expectant mothers - creating a closed loop between the patient at home and the care team in the clinic.
This episode covers what it takes to build health tech that actually gets adopted, why more data can create more liability before it creates better outcomes, how AI is entering the clinical documentation workflow first, and what the future looks like when every pregnancy has a continuous monitoring layer under it.
What Babyscripts Actually Does
Babyscripts is a B2B remote patient monitoring platform sold to health systems, OB-GYN practices, and nurse midwives. Providers prescribe the platform to pregnant patients, who receive a kit at home - typically a connected blood pressure cuff and cellular or Bluetooth-enabled devices - and a companion app that collects biometrics, symptoms, and behavioral data throughout all four trimesters.
The system monitors blood pressure, weight, gestational weight gain, blood sugar (for patients with gestational diabetes), and symptoms associated with preeclampsia - a dangerous high-blood-pressure condition in pregnancy. It also screens for social determinants of health: transportation access, mental health markers, depression, and suicidal ideation. If something requires immediate attention, both the patient and their care team are notified. The loop is complete without requiring an office visit.
The result is a shift from episodic, appointment-based care to continuous, asynchronous monitoring. Patients feel more empowered - they are active participants in their own data. Providers get escalation alerts for the risks they already know exist but have limited visibility into. And the system reduces unnecessary clinic visits while catching the critical ones that would otherwise be missed.
Frameworks from This Episode
These frameworks have been added to the AI for Founders Frameworks Library. Filter by Anish Sebastian (Babyscripts) to find them.
The Remote Monitoring Closed Loop
The value of remote patient monitoring is not just data collection - it is completing the loop. Collect biometrics at home, surface actionable alerts to the care team in real time, and close the loop with a clinical response before the patient ever needs to come in. Each break in the loop - data collected but not reviewed, alerts sent but not actioned, notifications delivered but not contextual - degrades the outcome. The loop only works when all three steps are tight.
The Data Liability Paradox
In healthcare, more data creates more liability before it creates better outcomes. A doctor who has access to a patient's sleep data, Oura ring readings, or continuous glucose monitor feed is exposed to information they may not have clinical protocols to act on - and in a litigious specialty like OB-GYN, that exposure matters. The implication for health tech founders: clinical validation of when to intervene (and when not to) is as valuable as the data itself. Publishing studies showing which signals require action and which are noise is product work, not academic work.
AI Documentation Before AI Diagnosis
The first wave of AI in clinical settings will not be diagnostic - it will be administrative. Every patient interaction needs to be transcribed, summarized, documented, and converted into a billing event. This is the highest-friction, lowest-value work in a physician's day. LLMs that automate clinical note generation, documentation, and billing are the fastest path to adoption because they reduce burden without requiring clinical validation of the AI's medical judgment. Build the time-saving layer first; build the diagnostic layer second.
Tools from This Episode
Remote patient monitoring platform for pregnancy. Ships connected devices (blood pressure cuffs, glucose monitors) to patients at home and creates a continuous monitoring loop between expectant mothers and their OB-GYN care teams.
This Week's Experiment
Map Your Patient Touchpoint Gaps: Where Does Monitoring Go Dark?
For any healthcare or high-touch service product, audit the gap between your customer touchpoints. List every scheduled interaction (appointments, check-ins, automated messages). Then list everything that happens to the customer between those touchpoints that your system cannot see. For each invisible gap, ask: what risk could escalate here without detection? What data point, if you had it, would change the clinical or business outcome? That gap map is your RPM product roadmap. The highest-value features are almost always in the spaces where your visibility goes dark.
Why Maternal Care Is Still Broken
The US maternal mortality rate is rising while every other advanced economy is declining or holding steady. Anish frames the core problem as access failure, not clinical failure. Patients who live far from their OB, who have transportation barriers, or who are in the weeks between appointments when a complication is quietly developing - those patients are invisible to the system until they show up in an emergency.
The payment structure compounds the problem. Fee-for-service healthcare pays physicians for encounters, not outcomes. A doctor who spends 20 minutes reviewing a patient's Oura ring data gets paid nothing for it. The incentive structure actively discourages the kind of continuous monitoring that would catch the problems fee-for-service cannot see. Value-based care - where a physician receives a flat fee to manage a patient's health over time - changes that math. Babyscripts is positioned as a tool that becomes far more valuable under value-based contracts.
How Two Technology Waves Changed the Company
Babyscripts was founded in 2013, when a telemedicine appointment was a novel concept and remote patient monitoring was largely unproven. The first several years were spent on clinical validation - publishing studies with academic medical centers, demonstrating that the platform actually changed outcomes, building the evidence base that conservative physicians require before adoption.
COVID was the first wave. Patients could not leave their homes; providers had to deliver care remotely. The adoption curve that would have taken a decade compressed into months. Since then, the pendulum has swung back somewhat as the healthcare system figures out the right balance of in-person and virtual care.
The second wave - AI - is arriving now and Anish believes it will be larger. The near-term application is administrative: LLMs generating clinical notes, summarizing patient interactions, and handling the documentation burden that currently pushes physicians to their laptops at 10pm. The longer-term application is predictive: risk algorithms trained on the longitudinal pregnancy data Babyscripts has accumulated, surfacing signals that are too subtle for a physician to catch manually but visible in aggregate to a well-trained model.
Q&A
Who is Babyscripts' customer?
Babyscripts sells B2B to health systems, OB-GYN practices, and nurse midwives. Providers then offer the platform to their patients, who use it under a doctor prescription or approval. The model is B2B2C: the paying customer is the provider, the end user is the expectant mother.
What data does Babyscripts collect?
Blood pressure (via connected cuff), weight and gestational weight gain, blood sugar (for gestational diabetes patients), symptoms associated with preeclampsia, social determinants of health (transportation access, food security), mental health markers including screening for depression and suicidal ideation, and education engagement data showing whether the patient is completing recommended care tasks. Future roadmap includes remote sonography and fetal monitoring.
Why does more data sometimes create more liability for physicians?
In a litigious specialty like OB-GYN, a physician who has access to a data point is potentially expected to act on it. If a patient shares sleep data from a wearable, that doctor now has information they may not have established protocols to act on - and if something goes wrong, a patient's attorney can argue the doctor saw warning signs and failed to respond. Until clinical protocols exist specifying when a given data point requires intervention, collecting it can expose physicians without improving their ability to act. This is why clinical validation studies - showing which signals are actionable and which are noise - are not just academic: they are product requirements.
What does AI unlock for Babyscripts specifically?
Two phases. First: administrative AI. Every patient interaction must be transcribed, summarized, documented, and converted to a billing event. LLMs can automate all of this, eliminating the documentation burden that currently occupies physicians after hours. This is the fastest near-term path to AI adoption in clinical settings because it reduces work without requiring clinical validation of the AI's medical judgment. Second: predictive AI. The longitudinal data from hundreds of thousands of pregnancies - biometrics, symptoms, outcomes - is a training set for risk prediction models that could identify escalating preeclampsia, gestational diabetes complications, or other high-risk patterns before they become emergencies.
What does Anish see as the future of care delivery?
A team of specialists available asynchronously on demand, coordinated by a primary care physician. Instead of finding a local dietician and scheduling weeks out, a patient would have access to a specialized agent or clinician through a single platform that maintains their longitudinal health record. Some of that team will be AI-driven; some will be human. The primary care physician acts as a coordinator. The result: a physician who today manages 3,000 patients could manage two to four times that number with AI augmentation. That scale is the only way to close the access gap.