The pitch sounds almost too elegant: strip out the administrative bloat from home care, hand those savings to underpaid caregivers and overstretched families, and let artificial intelligence handle the rest.
That's the wager behind Cova, a Y Combinator-backed startup that recently opened shop in Detroit. The company claims its "entire operation runs on AI agents"—a bold assertion in an industry where margins are razor-thin, turnover is brutal, and the work itself remains stubbornly analog. Whether automation can genuinely solve the economic riddle that's plagued home care for decades is, well, the half-million-dollar question.
According to its Y Combinator profile, Cova is promising Michigan caregivers competitive rates while simultaneously serving Medicaid patients, long-term care insurance clients, and private-pay families. If that sounds like squaring a circle, you're not alone in wondering how the math works.
Three Doors, One Promise
Cova's marketing materials offer three entry points, each framed around money. Existing caregivers are invited to "raise your rate" by switching agencies. Families hunting for care can "lower your rate and raise your caregiver's." And family members already providing unpaid care to loved ones? They can "get paid the highest rate" by enrolling through Cova's Medicaid pathway.
It's a tidy pitch. The execution, though, hinges on navigating Michigan's Home Help Program—a Medicaid-funded personal care initiative that allows relatives and friends to become paid caregivers. Anyone who's tried to thread that bureaucratic needle knows it's not simple.
Cova built an entire educational microsite, MichiganHomeHelp.com, to walk families through the process: CHAMPS enrollment, electronic visit verification logs, MDHHS deadlines, billing across Medicaid managed care plans and private insurance. "Cova handles it for you, for free," the site declares. The approval process can stretch for months; the company is betting that by smoothing out the friction, it can capture market share in a fragmented sector.
The AI Black Box
What exactly does "runs on AI agents" mean in practice? Cova hasn't said. Not publicly, anyway.
The company's Y Combinator listing makes the claim, but offers no specifics on which functions are automated or how deeply the technology penetrates daily operations. That ambiguity is familiar territory in the home care world right now. BAYADA, a much larger player, announced an "AI-enhanced home care model" in late 2025. UK-based Cera rolled out what it called "AI agents to automate home care" earlier this year. Software vendors like Caribou, CareReady AI, and Caregence have been pitching scheduling tools, intake automation, and operational dashboards to agencies looking for an edge.
The timing may not be coincidental. Michigan rolled out new electronic visit verification requirements for Medicaid managed care at the start of this year, the same month the state bumped up individual caregiver rates under the Home Help program. Compliance infrastructure is getting heavier just as reimbursement rates inch upward—a narrow window for efficiency plays, assuming the technology works as advertised.
Show Me the Money
Here's where things get murky. Cova's website asks "How much do caregivers get paid?" in its FAQ section, then doesn't answer the question.
One Reddit user posted mid-year that Cova quoted $22 an hour for an aunt enrolled in Medicaid. It's a single, unverified data point, but it's also the only publicly available wage figure attached to the company. For context, Michigan's agency-based home care market is crowded: Zentra Home Care, Entyre Care, MI Home Help Agency, Mission Home Help, and Tohme HomeCare all compete for the same Medicaid Home Help segment. Private-duty agencies like Relevar and Hayaat serve the cash-pay side. Cova is wading into a market where labor costs are the dominant expense and everyone is already squeezing margins.
If you're making competitive wage claims, transparency might help. The company hasn't offered it yet.
Early Innings
As of mid-2026, Cova lists a five-person team and was founded recently, according to its Y Combinator profile. Founders Jordan Ibe and Ansh Tandon worked with YC partner Jared Friedman. The company went through Y Combinator's accelerator program, receiving the standard deal structure: $500,000 total, split between $125,000 for 7 percent equity and $375,000 on an uncapped most-favored-nation SAFE.
The company's website displays a phone number—(734) 707-7330—and includes separate contact points for caregivers and families. Cova states it is licensed in Michigan, though public license verification wasn't completed for this story.
The site is live. The phone line is active. At least one person on Reddit has engaged with the onboarding process. Those are the tangible markers of a functioning business, however nascent.
The Unit Economics Puzzle
The underlying question is whether AI agents can actually compress operational costs enough to make premium caregiver wages viable in a Medicaid-reimbursed environment. Home care agencies typically spend 60 to 70 percent of revenue on direct labor, with another 20 to 25 percent on administrative overhead—intake, scheduling, compliance, billing, HR. If Cova can genuinely automate that second bucket without sacrificing quality, the model could work. If human overhead creeps back in—and it usually does—the economics collapse.
Michigan's regulatory environment just got more complex and slightly more lucrative at the same time. Cova is testing its hypothesis in that sliver of opportunity. Whether it scales will depend on whether the AI actually runs the operation, or whether the promise outpaces the reality.
For now, the startup is doing what startups do: making a big claim, opening for business, and seeing if anyone shows up. The home care industry has heard bold promises before. This one comes wrapped in code and backed by Silicon Valley. That doesn't make it wrong—but it doesn't make it proven, either.
