Kasumi Quinlan Is Using AI to Reimagine Food Security

Kasumi Quinlan co-founded Lemontree to solve food insecurity. There's no reason for people to go hungry when there's free food right around the corner.

Although there are more food pantries in the U.S. than there are McDonald’s, fewer than 1 in 4 food-insecure households take that help. That’s because the system is fragmented, confusing, and often stigmatizing. Families aren’t going hungry because the resources don’t exist. They’re going hungry because accessing those resources is too hard.

Operating in 11 regions across the country, including New York City, Atlanta, Tampa, Detroit, and New Jersey, Lemontree makes food resources more accessible for all. The AI-powered tool is centered on delivering a human experience, guiding families to the nearest free food options that best meet their needs.

In this interview, Kasumi describes how Lemontree built its AI model to feed families in need — without compromising on humanity.

How did your journey inspire you to explore AI for humanity?

Until 2020, our team was delivering meal kits to schools. We were passionate about food access, but the COVID lockdown upended everything. We had to figure out how to help our neighbors from afar.

The very first version of the Lemontree helpline rang to my cell phone, where I’d spend hours talking to neighbors looking for food resources. As more and more people started calling – not just for information, but for support – my team knew we had to find a way to scale the human aspect of our work.

We built Lemontree with the idea of operationalizing “hospitality” as much as we could. For us, that means using technology to automate the stuff computers do well, so that our frontline staff can focus their energy on preserving human connection in our work.

How is Lemontree using AI to connect people with access to food?

AI is the next generation of automation. But at Lemontree, our ethos is all about empowering our staff and our clients to bring their full selves into each interaction. AI saves us the time and energy to do that well, and at scale.

Navigating food assistance is nuanced and difficult. Lemontree personalizes the experience, so that families can find the best resources that match their needs. Building on our human decision-making process, we built an AI model that recommends a personalized set of food resources for each client. The model considers things like food pantry requirements (ID requirement, registration process, etc.), travel preferences, and the experiences of others at a particular food pantry — the kind of calculus a social worker might do, but for thousands of families daily.

For example, if Pantry A is around the corner, but it’s only open once a month and doesn’t offer great food, we’re more likely to recommend Pantry B, which is 10 minutes away, but is open every day of the week and has glowing reviews from your neighbors. But if that pantry hasn’t updated their website in a year, and those reviews are too old, we’ll tell you about Pantry C, which is a bit further, but easy to reach by public transportation (if that’s your travel preference).

The recent temporary disruption to SNAP benefits exposed just how precarious food security is in the U.S. As we continue to build community infrastructure that helps feed people in need, what role can AI play in closing the gap?

The first two weeks of November were unlike anything I had seen since we started this work in 2020. On Tuesday, November 4, Lemontree helped over 21,000 families access food. That was an all-time high for our organization, and up from 12,000 families just the previous Tuesday.

In the weeks leading up to November 1, we knew what might happen if families lost access to SNAP benefits. Our on-demand, elastic infrastructure is already designed to scale and meet fluctuating demand. It’s the only reason we were able to go from serving 12,000 families daily in one week, to serving 21,000 families the following week.

Even now that the shutdown is over, our team is uncovering where we can build even more resilience. We’re always working on technical solutions that decrease manual burdens and make us more efficient, as well as human solutions. A lot of the ideas we’re discussing aren’t new, but the urgency of the moment is forcing us to innovate more quickly. AI enables that rapid response.

How has Lemontree designed its AI product for scale?

For starters, we spent two years building out our data infrastructure before we even developed our resource recommendation model. We knew that we needed a reliable foundation to grow from. We also made deliberate choices about low-tech accessibility. Ultimately, our goal isn’t to have the most cutting-edge tech product. We exist to help every person who needs it. We use SMS and WhatsApp so families can get connected to food resources from their existing devices and preferred communication channels. Most importantly, we built a real-time feedback loop that continually improves our recommendations, based on the lived experience of every person we serve. That means our AI product only gets better as we get bigger.

What core values drive your unique vision for impact in an AI-driven future?

Honestly, I don’t think our future will be, or should be, driven by AI. It should be driven by us. Lemontree’s approach is rooted in empathy and hospitality, meaning we believe in solutions that feel joyful and that are informed by the people they serve. We want to build something that makes people feel seen and loved. So much of our world is designed around scarcity and fear. There is enough room for every human being to thrive. AI offers us new and faster ways to accomplish any kind of impact, but it’s our responsibility to use it wisely. To preserve and propagate what’s important to us.

Which visionary leaders, philosophies, or movements give you hope for a more human-centered AI future?

Actually, someone I heard speak at the recent Fund.AI event — Baratunde Thurston! He had such an inspiring way of talking about technology and the future that reminded me of Lemontree’s values. And he doesn’t come from a tech company; his background is in comedy and writing (or storytelling, as he put it). Bringing those different disciplines to the conversation on AI is really compelling. We’ve heard a lot about AI’s capacity to spin up bigger numbers, allowing us to access or achieve more of what we already have. But I’d like to hear more perspectives. In adrienne maree brown’s words: “We have the gift and responsibility to imagine.” That’s what makes us human.

What is your 7-word autobiography?

On this earth to do my part.