Thesis

We initially set out to build a hardware device, but after realizing how insanely long the FDA process would be, and how heavy the compliance, legal, and regulatory burdens were, we decided to pivot. We stayed in the nursing-home market because we had already built traction and relationships there, which was incredibly valuable. Our venture began with a very clear pain point: the enormous burden of 24/7 documentation in U.S. nursing homes. Nurses spend 80% of their time sitting in front of a computer or writing notes, and in about 30% of facilities documentation is still done by hand because they lack an EMR system. The problem is critical both medically (due to complex resident histories and years-long documentation) and financially (because insurance reimbursement depends entirely on accurate, comprehensive reporting).

Our new hypothesis was to build an AI documentation assistant that leveraged ambient listening of nurse-patient conversations to automatically, accurately, and quickly generate documentation directly into the nursing home’s EMR system.

The system would record all nurse-patient conversations (for example during vitals or medication delivery), perform transcription, extract key information, and automatically generate specific notes like progress notes inside the EMR.

Our broader goal was to enable simple input/output from these extremely difficult EMR systems, potentially through a chat/AI interface. We believed a tailored system aligned with nursing-home codes, forms, and repetitive workflows could meaningfully reduce the documentation burden.

Validation

We quickly realized that validation had to happen in the U.S. because the entire ecosystem – business model, operations, reimbursement – was completely different from what we knew in Israel, and this became a major point of failure.

Reaching nursing homes remotely was extremely difficult; cold calls barely converted, so to build real connections we spoke with four nursing-home chain owners (each managing at least ten facilities), administrators, and nurses. When facilities didn’t respond, we literally knocked on doors in New Jersey to meet people face-to-face. 

We also understood that implementation would require someone physically present and hands-on for onboarding and ongoing monthly visits. There is no clear playbook for selling to nursing homes, and our validation was shaped heavily by navigating this difficulty.

When we built a demo, nurses were amazed and told us they would use it in production, yet immediately added layers of constraints: it couldn’t record all the time, it had to extract only specific data, they didn’t want to use their phones due to privacy concerns, and more. That’s when we realized just how enormous the required market education would be.

Management, meanwhile, insisted the product had to be “really, really cheap” and repeatedly asked for clear ROI, which meant we needed a pilot to collect the data that would prove it.

We met several investors and mostly received the familiar feedback of “Show me A and B, and I’m in.” One investor with strong exposure to the sector (a doctor within a chain) was especially enthusiastic. Investors also consistently advised us to raise $1.5M instead of the $500K we planned.

Why it didn’t progress

1. Lack of CTO: The absence of a CTO was a critical red flag for investors, and we learned that it is better to establish the full founding team together rather than try to complete it later.

2. Market Inaccessibility: Trying to validate and sell in the U.S. while living elsewhere was the hardest challenge. In this domain, founders need to be physically close to the market from day one.

3. Lack of “Sexiness”: The nursing-home sector simply wasn’t considered “sexy,” making it extremely difficult to recruit people who shared our passion and long-term drive.

4. Market Education: Looking at AI capabilities today, we believe we were too early due to lack of market maturity. The market-education burden, which was enormous for us, would likely be far smaller now.

Insights

1. Misaligned Incentives Are Fatal: If end-users benefit but buyers don’t feel the value yet, the business collapses.

2. Market Access Is Non-Negotiable: Remote validation in this space creates endless loops of partial understanding; we learned not to build for a market we cannot physically access.

3. Opaque Markets Are Dangerous: Innovation in politically entangled or protectionist industries often threatens existing power structures.

4. CTO Parity Is Critical: The CTO must be treated as an equal, full partner from day one.

5. Vague Investor Feedback Means “No”: Conditional commitment is not a maybe. Real interest sounds like “I’m in.”

6. Permission to Fail Matters: Defining the venture as a learning journey from the start helped us avoid emotional burnout and co-founder friction. Failure should feel like progress, not collapse.

---