Why We Are Building Nobokai Care

The problem

People are living longer, and most want to age at home. The number of Americans aged 85 and over will double by 2040. The people who give that care are too few: home care loses about three in four caregivers every year.

Home care agencies carry this. They are small businesses doing some of the most important work there is, and most can’t afford to staff a phone line around the clock. When a daughter calls at 8pm about her mother and no one answers, she calls the next agency, or her mother goes without care. The staff who stay spend their days on phone calls, schedules, forms and state portals instead of on people.

Hiring alone can’t close this gap, because there aren’t enough people. What we can change is how much each person is able to do. Coding agents made engineers many times more productive. Care teams still run on phone tag and paperwork. We are building the agents for them: voice agents that answer every call and turn it into completed office work, inside the tools agencies already use, with a person approving anything that changes someone’s care.

Why this is personal

Tausif

Care has been the thread in what I build since I was a kid. At 14, I built an internet-connected pill dispenser for my grandma so she could take the right medicine at the right time. In high school I prototyped ProCure, a QR-based prescription system for patients, doctors and pharmacists in Bangladesh. It never went past a prototype, but it showed me that much of care breaks down in records and handoffs, not at the bedside.

Since then I’ve worked on assistive technology: smart skin patches in a miniature robotics lab at Vanderbilt, and a passive bedside monitor for elderly patients as my senior design project. I also built voice agents at the edge, including an open-source ESP32-S3 voice-agent board I designed end to end, now used in AI toys. When we saw how understaffed home care agencies are, everything I had worked on pointed at the same problem: voice agents, technology people depend on, and care.

Mashiur

At Vanderbilt, I built AI tools alongside nursing faculty, researchers, students and caregivers, with collaborations at Vanderbilt University Medical Center. Through regular consultations with nurses and nursing students, I saw that access to AI alone wasn’t enough. They needed tools that understood their work and followed rules they could trust. I coauthored two nursing AI papers and led AI coaching and workshops.

As a software engineer at Oracle, I worked on OCI infrastructure automation and healthcare projects. Working with home care agencies brought these experiences together: their teams needed tools that knew their people, rules and routines. That is what Nobokai Care is.

Why we are the team to solve it

We trust each other. We’ve known each other for more than a decade. We spent eight years at the same school in Dhaka, applied to US colleges together, and both came to Vanderbilt. We are both full-time, with a 50/50 equity split. Startups fail when founders stop trusting each other; ours was built long before the company.

We have already shipped together. Khanflow, the iOS calendar and to-do agent we built for fun, has more than 1,500 daily active users. Its core rule is the same one Nobokai Care runs on: it asks before it changes anything, and people can see everything it knows about them. We have already learned how to build an agent people trust with their day.

Our skills cover both halves of the product. Nobokai Care needs voice agents that hold up on real phone calls, and AI that care staff can trust to follow their rules. Tausif has built voice agents on low-power hardware and has worked on technology that has to work outside the lab, at Stanford AI Lab, Zipline, Oshkosh AeroTech and Synaptics. Mashiur has worked closely with nurses and caregivers, learning how they work and what they need from AI before they will trust it.

We both build end to end. Neither of us hands work off to the other to finish. What we learn from agencies goes straight into the product.

We are not chasing a trend. One of us has been building for care since he was 14; the other has been building AI alongside nurses at Vanderbilt. We would be working on this problem with or without a company.

What we still have to prove

We don’t have paying agencies yet. Our next steps are to get agencies live, show the calls we answer and the hours we give back, and add an advisor who has run a home care agency. We’ll be judged on outcomes: calls answered, shifts covered, visits paid, hours returned to care.

Our promise

Care is human work. Sitting with someone, noticing what has changed, comforting a family: that belongs to people. We will take the calls, schedules and forms off their plate, keep a person in charge of every decision about someone’s care, and say clearly when you’re talking to an AI.

We’ll know we’re getting there when no family’s call goes unanswered.

— Tausif Samin and Mashiur Khan