A multilingual AI voice platform connecting surplus food from restaurants and grocery stores to verified food shelters and community members in need. Built at HackDavis 2026 in 24 hours. Our number is (984) 302-3261.

1 in 7 Americans face food insecurity while 30% of grocery store food gets thrown away. The people who need food most are locked out of every existing solution — no smartphone, no data plan, no documentation, no English. We built something with zero barriers. One phone call. Any language. Done.

Donor flow: A restaurant calls one number and says "50 burritos ready at 9pm." Mesa's AI parses the listing, saves it to the database, and immediately calls nearby verified food banks to alert them. The food bank says yes, the listing is claimed, and the donor gets a callback confirming pickup.

Recipient flow: Anyone calls the same number in English or Spanish, gets connected to nearby shelters with available food, and finds out exactly where to go — no account, no ID, no app required.

Mesa is a fully agentic system. Nine tool handlers cover the complete flow: caller identification, donor registration, food bank registration, listing creation, outbound food bank notification, listing claims, organization verification, and nearby shelter lookup. The AI autonomously classifies donations, routes to the highest-need recipient, escalates as expiry approaches, and confirms pickup — no human dispatcher needed.

Getting Vapi's webhook integration right was the core technical challenge — the dispatcher has to respond within ~5 seconds with exactly the right payload format or Vapi rejects it. Parallel async checks for organization verification (EIN, geocode, web presence) had to be carefully time-bounded to stay within that window.

Devpost: https://devpost.com/software/mesa-3bxth0
Original repo: https://github.com/adamkai7/hackdavis26