The pitch from Cova sounds almost too tidy: automate away the administrative bloat that plagues home care agencies, redirect those savings to caregiver wages, and suddenly Michigan's elderly and disabled have access to better-paid workers without families footing a bigger bill. Whether reality cooperates with that vision is another question entirely.
But the Detroit startup—just five people, all told—is betting it can pull off something the industry hasn't managed in decades. Fresh from Y Combinator's recent cohort, Cova went live in Michigan this month as what it's calling an "AI-native home care agency." That's not marketing speak for a regular agency with some chatbots bolted on. Nearly every operational workflow, from intake to billing, runs through automation rather than human administrators.
The company claims to pay caregivers "the highest base rate in the state" while serving Medicaid patients, long-term care insurance holders, and private-pay families. That claim remains unverified—no published pay tables yet, and no independent confirmation. But the underlying bet is stark: use software to redistribute margin from back-office overhead to frontline workers. If it works, it's a model that could upend assumptions about what AI can actually accomplish in heavily regulated healthcare delivery.
Inside the automation
Co-founders Jordan Ibe and Ansh Tandon spent the past year building AI systems to navigate Michigan's bureaucratic thicket—MDHHS's CHAMPS enrollment portal, electronic visit verification logs through HHAeXchange, Medicaid billing for the state's Home Help program. The kind of paperwork that normally requires dedicated care coordinators and compliance staff.
"Our agents handle every workflow, from intake to benefits coordination," Ibe wrote in a LinkedIn post announcing the company's YC backing. Cova's website, along with an educational companion site called MichiganHomeHelp.com, walks families through eligibility requirements and program nuances. The sort of hand-holding that usually happens over the phone with someone who may or may not have time to explain the state's labyrinthine rules.
The operational model is deliberately narrow. Five employees. One state. Three payment models. It's less a blueprint for national scale than a controlled experiment: can AI genuinely run a licensed care agency end-to-end, or just smooth out the edges?
One Reddit user in Michigan mentioned working with Cova and cited a Medicaid caregiver rate of $22 per hour for their aunt—though that's anecdotal and the company hasn't officially confirmed its rate structure. For context, general market rates for caregivers on consumer platforms in Michigan hover around $19.91 per hour. Medicaid Home Help rates, though, operate under different reimbursement models—comparing them directly is tricky.
Why Michigan, why now
Cova is zeroing in on Michigan's Home Help program, which provides Medicaid-funded personal care for activities of daily living. The state bumped pay rates for individual caregivers in January, signaling at least some policy movement on the wage front. Perhaps creating an opening for an agency promising to push rates higher still.
The company accepts all three major payment sources: Michigan Medicaid, long-term care insurance, and private pay. That matters in a market where families often piece together funding from multiple sources as patients age or conditions shift. Cova's pitch is that it handles "enrollment, insurance, and billing"—taking those headaches off families and caregivers alike.
But the timing also reflects a broader industry moment. A January article in Home Health Care News noted that most at-home care providers want AI tools but cite cost as a barrier. A June newsletter reported that three-quarters of home care agencies are "hitting a wall with AI" despite enthusiasm. Translation: everyone wants the efficiency gains, few have figured out how to make the economics work.
Meanwhile, the federal government is circling the topic. HHS's Administration for Community Living launched a "Caregiver AI Prize" competition in February. The National Council on Aging released research in June outlining both promises and risks of AI in home- and community-based care. The interest is there. The execution, less so.
Cova isn't alone in the YC orbit, either. TakeCareOS, from the same batch, is building AI back-office tools for agencies—though as software sold to existing operators, not as a direct provider. Recent YC alums like Claim Health and Clara are exploring adjacent angles in senior care. Outside that ecosystem, Sensi.AI announced an "agentic operating system for home care" in June, and established players like WellSky and Homecare Homebase have published reports on AI integration at the point of care.
The difference is that most of those companies are selling picks and shovels. Cova is trying to mine the gold itself.
The harder test ahead
For now, the company remains Michigan-only. Founders Ibe and Tandon haven't publicly disclosed expansion plans or fundraising details beyond the standard YC backing. You can reach them at (734) 707-7330, they're operating from a Detroit address, and that's about all the public footprint they've established.
Which brings us to the real test. Automating workflows is one thing—doing it reliably enough to earn trust from vulnerable patients and their families is something else entirely. Home care agencies live and die on caregiver retention, patient satisfaction, and regulatory compliance. All three are notoriously hard to scale, and none forgive sloppiness.
If Cova can prove the model works—and if those caregiver wages genuinely stay competitive as the company grows—it could reshape what the industry assumes AI can handle. But it's a big if. Michigan's Home Help program serves thousands of families, and none of them signed up to be beta testers for an automation experiment, however well-intentioned.
The startup's founders seem to understand the stakes. In an industry where trust is earned one home visit at a time, being the most efficient agency won't mean much if you're not also the most reliable. Software can handle the paperwork. Whether it can handle the harder, messier parts of caregiving—that's still an open question.
