Baba Raises $40 Million to Put AI Care Teams Behind Medicare Patients, With 100,000 Families on Its Waitlist

Baba, an applied AI company building care navigation for Medicare and Medicaid patients, has raised a $40 million Series A led by FirstMark Capital, with CVS Health Ventures and value-based care investor Valtruis joining returning backers General Catalyst, Genius Ventures and Ground Up Ventures. The New York company announced the round on October 20, and Fierce Healthcare reported on it this weekend.

Founded in 2025, Baba pairs every enrolled patient with a dedicated human care team of care coordinators and licensed therapists, supported by AI agents that handle the administrative layer of American healthcare: specialist referrals, hospital-to-home discharge tracking, in-home aide coordination, billing denials, and applications for housing assistance and medical transportation. The pitch is that navigating care for the nearly 70 million Americans on Medicare has become a family member's unpaid second job, with more than 63 million family caregivers nationally, up 20 million over the past decade.

The operating numbers, all self-reported and not independently audited, show a demand gap that explains the round. Baba says it has supported more than 10,000 patients and families across the US, with more than 100,000 families on its waiting list, a roughly ten-to-one gap between demand and current capacity. Founder and CEO Connor Sweeney framed the AI layer as a cost argument: by dramatically lowering the cost of operating and coordinating care, care delivery becomes economically viable for populations that were historically too expensive to serve.

The distribution asset is a government one. In June, the Centers for Medicare and Medicaid Services selected Baba for the ACCESS Model, the CMS Innovation Center's ten-year test of technology-supported chronic care, across all four of its clinical tracks: early cardio-kidney-metabolic, advanced cardio-kidney-metabolic, musculoskeletal and behavioral health. Value-based arrangements tie revenue to verified outcomes rather than fee-for-service billing, which is why a care navigation startup can raise at all.

Baba is also trying to make its AI claims testable. Alongside the round it released BabaBench, a benchmark measuring whether frontier AI agents can complete long-horizon patient advocacy tasks end to end, paperwork, calls with insurers, billing appeals and follow-through that stretches across weeks or months, without privacy violations or prohibited clinical actions. The benchmark is posted as an arXiv preprint and accepted to an upcoming NeurIPS workshop, a more rigorous posture than most startups' marketing benchmarks.

The strategic investor is the signal worth watching. CVS Health Ventures is the venture arm of CVS Health, which owns Aetna, one of the largest Medicare Advantage insurers, so its participation puts a direct payer counterparty on the cap table to test whether AI-assisted navigation actually lowers claims costs. FirstMark co-founder Rick Heitzmann called Baba a bet on the consumerization of healthcare for millions of Americans who are rural, older and have more complex needs.

Caveats apply. The round was announced via a company press release, and the patient, waitlist and caregiver figures all come from Baba itself. The CMS ACCESS Model is a ten-year demonstration, not a guaranteed reimbursement line, and CMS test programs have been narrowed or ended before. Competitors including Abridge, Commure and Sword Health are chasing adjacent pieces of the AI-plus-clinician care thesis, and a $40 million Series A is small against the staffing cost of clinician-led care teams.

The interesting question Baba raises for the broader agent market is scope. While consumer AI agents stumble on restaurant reservations, Baba argues that general-purpose agents cannot navigate healthcare at all, and that the viable design keeps licensed humans in the loop and trains on narrow, long-horizon bureaucratic workflows instead. If the model works, the lesson for AI agents everywhere would be unfashionable but familiar: the money is in the boring paperwork, not the demo.