There is a specific kind of business problem that sounds absurd until you say it out loud slowly, and then it sounds obvious, and then you cannot believe nobody fixed it. Home care is one of those. A caregiver visits a senior for a few hours, writes some notes, and leaves. For the other twenty or so hours a day, the industry's official position on what is happening in that home is: we have no idea. This is not a scandal so much as an accepted fact. Home care, as Sensi.AI likes to put it, has been flying blind for decades.
Romi Gubes, a former enterprise software engineer with a decade at companies like Cisco, Vonage, and DellEMC, looked at that gap and proposed something that made people uneasy: put a small, camera-free device in the home, let it listen to ambient sound, and use AI to turn that sound into care intelligence - falls, a cough that will not quit, the particular quiet of someone who has stopped talking. The reaction, by her own account, was skepticism. A listening device for the elderly is the sort of idea that reads as dystopian in the pitch and humane in the demo, and the whole trick of the company is getting you from the first reaction to the second.
She built it anyway, co-founding Sensi.AI in 2018 with Nevo Elmalem and Alon Brener. The reframe they landed on is not surveillance but presence: the device is there in the hours a human cannot be, and it does not judge, it flags. When it hears something that matters, it captures the relevant clip, produces a transcript, and sends the care team an alert with a recommended action. The senior gets watched over. The agency gets to see the hours it was previously guessing about.
What makes this more than a gadget is where it points. The senior care technology market is enormous - Sensi cites a figure around $390 billion - and it runs, to a startling degree, on paper, phone calls, and hope. That is the kind of market where you do not want to sell another app that agencies have to remember to open. You want to sell intelligence that layers onto work they already do. Sensi's bet is vertical, not general: it did not build a chatbot that can do a little of everything, it built a care agent that understands what a caregiver's actual day looks like, which is a harder and narrower and more defensible thing.
