A Boston startup put an AI voice agent on the phone and a licensed nurse behind it. The engine is called Glia. The bet is that America's nursing shortage is a capacity problem you can multiply your way out of.
Here is a fact about healthcare that everyone agrees on and nobody has solved: there are not enough nurses, there will not be enough nurses, and the phone keeps ringing anyway. Every projection for the next decade points the same direction - demand up, supply flat. You cannot conjure a licensed RN out of a job posting, and you certainly cannot conjure one out of a spreadsheet.
So the usual move in software is to look at that gap and say: fine, we'll automate the nurse. Build a chatbot, call it triage, ship it. This is the move that makes hospital compliance officers reach for the aspirin, because a wrong answer in nursing is not a wrong answer in, say, a food-delivery app. It's a person who was told their chest pain was probably heartburn.
OutcomesAI is making a different, more interesting move. Its pitch, stated plainly by founder and CEO Kuldeep Singh Rajput, is that the company is not replacing nurses - it is multiplying them. The distinction sounds like marketing until you look at how the product is actually built, and then it starts to look like the whole thesis.
Healthcare is running out of nursing capacity, and incremental fixes won't solve it. With OutcomesAI, we're not replacing nurses - we're multiplying them.
Figures are company-reported. Capacity and cost claims reflect early deployments, not audited results.
The company's core product is named Glia - a nod, if you want to read into it, to glial cells, the support cells that surround and assist neurons. The neurons in this metaphor are the nurses. Glia is the tissue that lets them reach further.
Mechanically, Glia does two things at once. First, AI voice agents pick up the routine calls - inbound and outbound - and handle the parts that don't require a clinical license: capturing symptoms, scheduling visits, coordinating follow-ups, delivering patient education, all in multiple languages. Second, when a call needs a human, a licensed OutcomesAI nurse steps in, now armed with AI productivity tools: real-time scribing, protocol guidance, and triage running on the established Schmitt-Thompson protocols that call centers already trust.
The elegant part is the division of labor. The routine, high-volume, low-judgment work - which is most of the work - flows to the machine. The moments that require a human flow to the human, who is no longer buried in the routine stuff. That's where the "3 to 5x more capacity" number is supposed to come from. Not magic. Just triage of the nurses' own time.
Glia voice agent takes the call, in the patient's language.
Structured intake, scheduling, education - the routine load.
A licensed nurse steps in when a case needs one.
Real-time scribing and protocol guidance behind every RN.
Manage inbound and outbound calls - symptom capture, scheduling, follow-ups, and multilingual patient education.
Symptom assessment and escalation on Schmitt-Thompson protocols, with AI-enabled licensed nurses in the loop.
Scheduling, referrals, and care navigation - the administrative load lifted off clinical teams.
Follow-up coordination and continuity management after discharge or an acute episode.
Kuldeep Singh Rajput did not arrive at nursing capacity by accident. He previously founded Biofourmis, one of the better-known names in digital therapeutics and remote patient monitoring - a company built on the idea of predicting and preventing serious medical events with data. He could, after that, have built almost anything. He chose the triage phone line.
This is worth pausing on, because the phone line is deeply unglamorous. There is no flashy demo in a scheduling reminder. But ask any nurse where the hours actually go, and it isn't the hard cases - it's the volume of routine contact. The follow-up call. The post-discharge check-in. The "I have a question about my medication." Starting there, at the boring bottleneck, is a choice that reads as operator instinct rather than founder theater.
White space in healthcare is rarely empty. It's guarded - by regulation, by safety, by liability. The graveyard of health-tech startups is full of companies that had a clever model and no answer for the question "but is it safe?"
OutcomesAI's answer was to build the answer first. In 2024, before scaling commercially, the company launched the OutcomesAI Collaboratory: five leading health systems and virtual care companies brought together for the specific purpose of validating Glia's safety, accuracy, and clinical performance. Early results, per the company, point to meaningful operational gains - hundreds of nurse hours saved per month, patient-to-nurse ratios doubled.
The sequencing tells you something about how this company thinks. Validate, then scale. In consumer software that order is a luxury. In nursing it is closer to a requirement, and building it into the product from day one is the sort of unglamorous decision that doesn't make a headline but does make a durable business.
Illustrative bars scaled from company-reported ranges. Not independently audited.
In October 2025, OutcomesAI announced $10 million in seed financing led by Santé Ventures, a firm with a long track record in health and life-science bets. The round came with a board that is worth reading as a signal. Joining founder Rajput were Joe Cunningham, M.D. and Dennis McWilliams of Santé Ventures; Linda Finkel, senior advisor at AVIA; and Kevin White, Ph.D., co-founder of Tempus AI - one of the largest names in healthcare AI.
When operators who have built at genuine scale show up for a seed round, it usually means two things: the problem is real, and the timing feels right. It does not guarantee the company works. It does suggest the people writing the checks think the phone-line thesis is more than a slide.
Led the $10M seed. Partners Joe Cunningham and Dennis McWilliams joined the board.
Co-founder of Tempus AI, joined the board - a direct line to healthcare AI at scale.
Senior advisor at AVIA; joined the board. Glia is also listed on the AVIA marketplace.
Strip away the funding narrative and the practical question is: who benefits, and how? For a health system, OutcomesAI is a way to absorb call volume without hiring a call center you can't staff, and - if the numbers hold - to do it for roughly half what outsourced triage costs. For a virtual care provider, it's capacity on demand. For pharma, it's a channel for patient support programs. And for patients, the promise is prosaic but real: the phone gets answered, in your language, faster.
OutcomesAI launches the Collaboratory - five health systems and virtual care companies validate Glia's safety, accuracy, and clinical performance.
Early deployments report doubled patient-to-nurse ratios and hundreds of nurse hours saved per month.
$10M seed led by Santé Ventures. Board expands to include the co-founder of Tempus AI and an AVIA senior advisor.
"Glia" comes from glial cells - the support cells around neurons. The nurses are the neurons.
Rajput previously founded Biofourmis, a leading digital therapeutics and remote-monitoring company.
The company built a five-system validation program before selling broadly - unusual sequencing.
Glia's voice agents work across languages, aimed at multilingual patient populations.