The product is the bridge. The data is the moat.
A recap of our April 21 session on AI as infrastructure — with Lucy Kosturko of Social Cascade
Thank you to everyone who joined us last Tuesday for our latest community session. We spent the hour pressure-testing a framework we keep coming back to when we evaluate AI-enabled companies in health, wealth, and climate: the distinction between the bridge a company builds and the moat it accumulates.
The bridge is the product — the thing that gets a user from one side of a broken system to the other. The moat is what compounds underneath it while users are crossing: the proprietary data, the trusted network, the performance signal that no one else can buy, borrow, or scrape. Great AI-enabled businesses in underserved markets tend to build both, and the second one is what actually becomes defensible. Or, as Lucy put it back to us on the call: "The product was the vehicle, but the data was the moat."
The case study: Social Cascade
Lucy is an AI researcher turned founder who did not set out to build a healthcare company. She set out to fix early literacy — and followed the problem upstream until she landed at pediatricians, the trusted voice that touches nearly every young child in America at least once a year. The gap she found: providers desperately want to show up for patients between office visits, but almost never have the time or infrastructure to do it.
Social Cascade closes that gap. They use AI not to generate content, but to optimize delivery — putting trusted, physician-led health content on providers' own social channels at scale. The traction speaks for itself: near-perfect retention, a provider network spanning most of the country, and a deep bench of national and local content partners feeding the system.
Where the moat lives
The bridge is real — Social Cascade is the mechanism that lets a small practice show up like a health system. But the moat is what we kept circling back to. Every time a new provider joins the network, Social Cascade learns something that no large incumbent can replicate: how the same piece of content performs differently when it comes through a trusted local voice — by geography, demographic, format, season, reading level, even local pollen count. Years of granular performance data across a distributed network of trusted voices that AI, on its own, cannot manufacture.
That is the kind of moat we get excited about. It is not a better model. It is a data asset that compounds with every new node on the network — and it is trusted-voice agnostic, meaning the same infrastructure could extend to K-12 educators or any other community figure who rarely shows up online.
The takeaway for founders and investors
Three questions we are now asking every AI-enabled company in our pipeline:
- What is the bridge — the product that gets users across a broken system?
- What is the moat — what compounds underneath, and what can AI not buy or build?
- Who is the trusted voice, and how are you earning the right to amplify it?
One enterprise customer ran Social Cascade as the only lever against primary care visits and saw a steady month-over-month lift in provider utilization. That is the pattern we are looking for: a product that moves a system-level metric because the data underneath it has already been doing the quiet work.
Thanks for spending time with us digging into the details of our thesis with this case study. We look forward to continuing the conversation with many of you in person!
In the meantime, if you are building in this space — or know someone who is — we would love to hear from you.
Best,
Sydney and Shruti
General Partners
Symphonic Capital