Check food access in your neighborhood. Enter your zip code to see if your area has limited access to fresh, healthy food.
Picture a grandmother in rural Kentucky. She qualifies for SNAP, WIC referrals, and food pantry access. She doesn't know about any of them. Not because the information doesn't exist, but because finding it requires navigating a maze of federal, state, and local systems that weren't designed to talk to each other.
She's not alone. 47 million Americans are food insecure right now. Millions of them qualify for benefits they'll never receive. The programs exist. The funding exists. But the path from "you're eligible" to "you're enrolled" is so fragmented that people just... give up.
That's not a knowledge problem. It's a navigation problem. And it's the problem we built FreshFind to solve.
We spent two years researching food insecurity in five very different U.S. communities. The same pattern kept showing up everywhere: the programs are there, but the barriers to actually accessing them are enormous. In Illinois, half of eligible elderly residents don't receive SNAP. Half. In Wolfe County, Kentucky, food pantries provide 1 meal for every 9 that SNAP delivers. The county's food insecurity rate is 27.3%.
And it's getting worse. The 2025 House Budget Plan introduced the largest proposed SNAP cuts in history and expanded work requirements to age 64. That puts 135,000 more Kentuckians at risk of losing benefits. The people most affected by these changes are the ones with the fewest resources to figure out what the changes even mean for them.
Think of FreshFind as the caseworker you wish existed. But one that's available 24/7, knows every program in your area, and can walk you through the entire process from "do I qualify?" to "your application is submitted."
We built this infrastructure on top of tools we already had in the field. It works with local governments, not around them. FreshFind wraps these foundational tools into a generative AI layer — a conversational interface that walks users through the system in plain language, the way a knowledgeable caseworker would.
The multi-step work is handled by agentic AI that acts on the user's behalf without requiring them to understand the underlying complexity. So it does four specific things:
You answer a few questions. FreshFind checks everything at once: SNAP, WIC, free school meals, food pantries, utility assistance, Medicaid nutrition programs. Instead of applying to six different places and hoping for the best, you get one clear picture of every resource available to you.
Knowing you qualify doesn't put groceries on the table. So FreshFind goes further: it finds the nearest stores that accept SNAP and actually carry fresh produce, pulls up food pantry hours, figures out how you'd get there, and checks what's in season. All based on your exact location.
This is where the real impact happens. Most people drop off during the application process because it's confusing, the forms are long, and nobody tells you what documents you need until you've already waited in line. FreshFind pre-fills what it can, walks you through the rest via a simple text interface, flags deadlines, and sends reminders when it's time to recertify. It's not a frustrating portal—it's a helpful digital caseworker right on your phone.
Policy changes constantly. New work requirements, benefit reductions, rule changes at the state level. Most people find out they've lost benefits after it happens. FreshFind monitors all of it and proactively reaches out to affected households: here's what changed, here's what it means for you, here's what to do next.
We didn't just read about food insecurity. In collaboration with the University of Virginia, we went to five very different places and mapped what was actually happening on the ground: who's going hungry, why the existing programs aren't reaching them, what's blocking access. That research is what FreshFind's AI actually runs on.
We don't ask people to download another app or visit another website. FreshFind plugs into places where people are already showing up. The goal is to meet folks where they are, not where we wish they were.
Families are already there for WIC and nutrition visits. Why not check everything else while they're at it?
In rural areas especially, the library is the one place everyone trusts and everyone can get to
If a kid qualifies for free lunch, the family probably qualifies for a lot more
People already call 211 for help. FreshFind makes that help more specific and more complete
Here's what government partners actually get: fewer incomplete applications clogging the system, more people enrolled, and real data on what their community actually needs. Caseworkers spend less time on intake and more time on the work that requires a human.
FreshFind doesn't sell or retain user data. We operate with strict privacy controls, complying with state and federal data guidelines (including HIPAA/PII protocols where required), and integrate securely with the systems you already rely on.
That's the question we asked ourselves before we started. So we deliberately picked five communities that have almost nothing in common: a major metro (Chicago), a high-inequality city (Atlanta), a mid-size metro (Kansas City), a growing rural hub (Twin Falls), and a deep-rural Appalachian county (Wolfe County, KY). The same approach worked in all of them.
Here's why it scales for governments: FreshFind is built as a replicable toolkit, not a custom project. Any state SNAP agency, county health department, or city 211 system can deploy it by supplying their local data — which stores accept SNAP, where food pantries are, how transit works, what the state rules are. The AI builds a local model on the same core infrastructure. No rebuilding. No new codebase. Just a new deployment. One government's success becomes another government's starting point.
And the model travels. FinMango has operated in 13+ countries, and the government navigation problem is universal. India's Public Distribution System, Brazil's Bolsa Família — the FreshFind architecture is designed so that any government running a social benefits program can replicate it. At the end of the 18-month grant period, we'll publish a full open-source toolkit so any city, state, or national government can deploy this without starting from scratch.
FinMango was founded in 2017 at Kent State. The original idea was simple: build free tools that help people navigate financial systems that weren't designed with them in mind. That idea turned into a 501(c)(3) nonprofit that's now reached over 1 million people in 13+ countries.
Along the way, our team started publishing original research on financial inequality, built the Food Assistance Calculator and Food Desert Analyzer, and grew a network of student ambassadors. We've partnered with Google Health, the World Health Organization, the World Bank, and the IMF. But the core lesson we keep coming back to is this: the tools are only useful if people can actually find and use them.
FreshFind is the most ambitious thing our organization has ever tried to build. It takes everything we've learned — the research, the tools, the community relationships — and puts AI to work on the specific problem of getting food assistance to people who need it. Not studying the problem. Solving it.
The Google.org Impact Challenge awards grants of $1M–$3M to projects using AI for government innovation. We're requesting $1.5M to take FreshFind from working prototypes to a deployed, scaled product over 18 months — and we're looking for 2–3 forward-thinking government partners to pilot with us.
Billions in food assistance goes unclaimed every year. Not because people don't need it. Because finding it is just too hard. We can fix that.
Scott Glasgow · Sarah Cherian · Soham Patel · Tony Ramos
FinMango · finmango.org
