Thesis

My project was driven by a specific observation of what I call the “Data Intake Gap” in the hospitality industry. While modern restaurants generate massive amounts of data through Point of Sale (POS) systems, the people running them are often non-technical and lack the time or analytical skills to make sense of it all.

I noticed that traditional analytics platforms force users to sit at a computer, navigate complex dashboards, and manually apply filters just to see basic performance metrics. For a restaurant owner constantly on the move, this is a major friction point. I believed that Generative AI could eliminate the need for a traditional UI entirely.

My thesis was that a WhatsApp-based “Chat Analyst” could serve as a virtual partner. Instead of hunting for reports, an owner could simply text: “How many chickens do I need to order for Monday?” or “What was my labor cost percentage last night?” and receive a precise answer instantly. My goal wasn’t just to fetch data, but to provide context. By connecting disparate sources (like scheduling software, inventory systems, and POS data) the AI could offer cross-functional insights, such as advising an owner to increase staff for a Monday shift because historical data showed a consistent spike in revenue on that specific day.

Validation

I approached the exploration with a highly structured methodology to ensure I wasn’t just building a “nice-to-have” tool. Initially, I didn’t just talk to restaurant owners; I segmented the entire hospitality software ecosystem. I conducted deep-dive interviews with 30-40 software providers (the “Platforms”) and 10-15 restaurant owners (the “End Users”) to understand where the information bottleneck truly lived, as well as the usage capacity of the platforms. Through these conversations, I discovered that owners didn’t necessarily want to “chat” all day. They wanted a proactive “Morning Briefing”. This led to a shift in my product vision: a push-notification summary delivered every morning that distilled the previous night’s complexity into five key KPIs to determine if the business was healthy. Restaurant owners expressed a clear intent to pay approximately $50 per month, per location, for this simplified access to their data.

One of the most successful parts of my exploration was testing the “Trust Barrier”. I found the pain was so acute that owners were willing to grant full admin API access to their live POS data (Toast/Tabit) almost immediately. I was able to move from a first meeting to having a live data sync in less than 24 hours.

Regarding the ICP, I initially looked at “Mom and Pop” shops, but the exploration stage quickly taught me that the real ROI was in mid-sized networks (5-20 locations). In these businesses, the owner is no longer physically present at every location and is therefore entirely dependent on data to understand what is happening on the ground.

We were highly active in the fundraising circuit, meeting with the major early-stage Israeli VCs and the majority of early-stage VCs in New York. While we raised approximately $150,000 from angel investors and nearly closed a round with a small venture fund after a long due diligence process, the deal fell through at the last moment.

Why it didn’t progress

1. The Go-To-Market (GTM) Trap: I realized that selling directly to restaurants (B2B) is a “trauma-inducing” grind characterized by high churn and low ACV. Conversely, trying to sell through POS providers (B2B2B) resulted in glacial sales cycles. POS companies liked the idea but didn’t view it as a top priority for their own revenue lines.

2. The “Wrapper” Risk: Investors raised a critical concern: was this a defensible company or just a “thin wrapper” on OpenAI? As OpenAI’s capabilities grew, the fear that the platform would simply “eat” this niche became a major fundraising hurdle.

3. Founder Reality Check: The “Perfect Storm” of personal life intervened. Between relocation, regional conflict, and critical personal matters, I realized I no longer had the bandwidth for the 24/7 obsession a startup demands.

Insights

1. Build Relationships with VCs Early:  We decided early on to wait until we had a working MVP and paying customers before approaching major VCs. In hindsight, I should have started building relationships with leading funds much earlier.

2. Solve the Revenue Line, Not Just the Headache: Analytics are a “nice-to-have” in a low-margin industry like restaurants. If your tool doesn’t directly increase the daily revenue line, it will never be a Top 5 priority for a POS provider.

3. Be Wary of “Vertical AI” Hype: Building a business on top of LLMs requires a deep “data moat.” Simply making data easier to talk to isn’t enough if the underlying platform can build the same feature as a weekend project.

4. Know Your Season: There is a time to be a founder and a time to be a parent. Admitting that you cannot give a startup 100% is not a failureת it’s a professional necessity to avoid a slow, expensive collapse.

---