A Phone Call That Reads the Label.
It's a Tuesday morning in a community oncology clinic somewhere between Tulsa and the rest of the country. A nurse practitioner has six minutes between rooms and a question about a new immunotherapy that her drug rep won't be visiting for another three weeks. She dials a number on a sticker. A voice picks up. It is patient, specific, and weirdly well-read. It cites the FDA label. It knows which adverse-event grading scale she means when she says "grade 3." It does not, at any point, sound like it is selling her anything.
That voice is Synthio Labs. It is also the reason a handful of the world's largest pharmaceutical companies have started writing the company checks while Synthio's headcount still fits in a single Slack channel.
The startup, founded in 2024 and tucked into San Francisco, is building what it calls "the intelligence layer behind modern life sciences engagement." Translation, for the rest of us: a voice AI operating system that can speak fluent pharma without breaking a single rule.
Why Pharma Needed Its Own Voice AI.
Most voice AI you've met was raised on customer service tickets and pizza orders. Helpful, sure. Cleared for medical conversations? Not even close. The life sciences industry runs on a tightrope: every claim is regulated, every off-label whisper is a compliance event, every promotional word is reviewed by a human with a red pen and a legal degree.
So when a pharma marketer says "let's add an AI agent," what they actually mean is "let's add an AI agent that has read the label, understands the safety profile, refuses to improvise, and logs everything." That is a different animal.
Synthio Labs's founders - Supreet Deshpande, Sahitya Sridhar, and Rajashekar Vasantha - came at the problem from inside the tent. McKinsey. ZS. Amazon AI research. They had watched pharma teams pay millions for engagement that didn't scale and engagement tech that didn't engage.
Their pitch, more or less: build the conversational layer from scratch with regulation as a feature, not a settings panel. Then plug it into the systems pharma already trusts - Veeva, Salesforce, medical information desks - so nobody has to migrate anything to use it.
Five Products, One Operating System.
Synthio doesn't sell a chatbot. It sells a small cast of specialists, each pointed at a different part of the pharma conversation.
Jarvis
A clinical-grade voice copilot for pharma field teams. Briefs reps, fields questions, never goes off-label.
Ather
Multimodal voice agents that engage healthcare professionals across phone, web, chat, QR, and SMS.
Helix
Voice AI for patient support - adherence calls, education, the long, unglamorous middle of a treatment journey.
Simulation Studio
High-fidelity digital twins of clinicians and patients, so strategy teams can pressure-test messaging before a single rep gets on a plane.
Polaris HQ
"Talk to your data." A data intelligence layer that turns SQL gymnastics into a sentence.
Where Synthio Plugs In
Relative product breadth across pharma workflows. Approximate, illustrative.
Compliance Isn't a Slide. It's the Architecture.
Most AI companies bolt safety onto a model after the demo lands. Synthio's bet is that the bolt-on era is over. The product ingests FDA labels dynamically. It runs answers through a clinical reasoning engine. It logs the chain. And it keeps a knowledge graph of approved claims so it knows what it's allowed to say before it says it.
That last detail matters. In pharma, the question isn't "can the AI answer?" - it's "is the AI allowed to answer this way, in this market, to this audience, for this molecule, today?" Synthio built the company around that question.
The compliance posture isn't a marketing line either - it lives at compliance.synthiolabs.com. When the trust center is a subdomain, it tends to mean someone is taking it seriously.
Pinned to the wall
- Founded2024 · San Francisco, California
- YC batchSpring 2025
- Seed$5M · Nov 2025 · led by Elevation Capital
- Co-investors1984 Ventures, Peak XV, Pioneer, Liama, YC
- FoundersSupreet Deshpande, Sahitya Sridhar, Rajashekar Vasantha
- CustomersMultiple Top-10 pharma + leading D2C health brands
The Short Trip From Demo Day to Top-10 Pharma.
Synthio Labs founded in San Francisco by Deshpande, Sridhar, and Vasantha.
Joins the Y Combinator Spring 2025 batch.
Onboards first Top-10 pharma customer; launches Jarvis, Ather, Helix, and Simulation Studio.
Closes $5M seed led by Elevation Capital with 1984 Ventures, Peak XV, Pioneer, Liama Ventures, and YC.
Demonstrates compliant HCP engagement at ASCO 2026, the world's largest oncology conference.
What People Actually Do With It.
A medical-affairs lead at a global pharma uses Ather to staff a 24/7 voice line that handles HCP medical inquiries in five languages without putting a human on overnight rotation. The line is fast. The line is compliant. The line, importantly, does not improvise.
A field sales director deploys Jarvis on a rep's phone, then runs the playback as coaching tape. The model flags when a rep wandered off-label, where they missed a competitive question, what the doctor actually wanted to know.
A patient-support program leans on Helix to call back people who didn't pick up their refill. Helix is gentler about it than the human team, partly because it has nothing else to do that day, and partly because it is reading from a script lawyers already cleared.
A strategy team runs Simulation Studio against a hundred synthetic oncologists before its real advisory board even gets the invitation. By the time the real KOLs walk in, the worst version of the meeting has already been held - and learned from.
Who Wrote the Check.
The November 2025 seed - $5M - was led by Elevation Capital. Joining the round: 1984 Ventures, Peak XV Partners, Y Combinator, Pioneer, Liama Ventures, and a roster of strategic angels from across global healthcare and AI. The check, in seed terms, is not unusual. The pattern of who signed it - growth-stage names taking a seed bet alongside YC - is.
Where to Find Them.
Back to Tuesday Morning in Tulsa.
The nurse practitioner hangs up. Her six minutes are nearly up. She did not get a brochure, a follow-up email cadence, or a sales pitch. She got the answer to her question, sourced and citable, in language a clinician uses with another clinician. The patient in Room 3 is waiting, and the line behind that call - the long, dull, unglamorous infrastructure that made it work - is invisible to her, which is how she'd prefer it.
Somewhere in San Francisco, in a Slack channel that still fits 23 people, a log entry rolls in. A timestamp. A molecule. A market. A claim used. A compliance flag, green. The work is mostly bookkeeping, and the bookkeeping is mostly the point.
That is the bet Synthio Labs is making. Not that voice AI will be everywhere - that part feels safe enough. But that the version pharma can actually deploy, the kind that reads the label and respects the rules and shows its homework, is a different category of product than the one the rest of the industry is shipping. And that category needs an operating system. The company is busy building it.