Have a question about starting a food business in Minnesota? Ask AVA.

Have a question about the food industry? Ask AVA.

For example, if you want to sell your family-favorite cookies to the farmers market, AVA says: “Register as a cottage food producer. This registration can be done online through the Minnesota Department of Agriculture website.”

And don’t forget: “The booth should be attractive and eye-catching to attract customers.”

AVA, a chatbot built by Minnesota’s Agricultural Utilization Research Institute (AURI), launched earlier this month to connect more food founders and ag producers to the nonprofit’s wealth of resources.

“With the rise of the cottage food industry, we’re seeing an increase in demand for our services,” said Jason Robinson, director of business development for food at AURI. But with limited staff to respond to those requests, “AVA represents an opportunity to connect those people to a knowledge base that’s curated by AURI experts but powered by generative AI to produce the answers that they’re looking for.”

The artificial intelligence-equipped resource is free for food entrepreneurs, farmers and anyone who has a question about the business of food. Similar to ChatGPT, Copilot and other AI tools, users can ask a question at auri.org/ava to prompt a computer-generated response.

Unlike other large language models, however, AVA only bases its responses on what AURI has provided. Popular AI chatbots and Google search summaries built on the entirety of the internet often suffer from “hallucinations” and provide wildly inaccurate results.

Not so with AVA, which stands for AURI’s Virtual Assistant, Robinson said.

“AVA is going to blend generative AI with the power of that human-centric library to really create answers that you can you can trust,” he said. “You’re not going to confuse food philosophy and food science, for instance.”

Users also can ask questions about their specific business plans without fear of seeing their trade secrets seep into answers AVA gives to others. Whereas ChatGPT learns from questions it is asked — and repeats information from those questions — AVA relies only on the material AURI feeds it.

“We were very purposeful in how we designed AVA in that way,” Robinson said. “It’s going to be a little slower in terms of how it learns but more trustworthy.”

He stressed that users not seeing an answer they need should tag the response and send feedback to AURI.

The U.S. Department of Agriculture paid for the project through a rural development grant.

Minnesota’s food and agriculture sector brings in $106 billion to the state economy and provides 388,000 jobs, according to the most recent data from the Department of Employment and Economic Development, a marked increase from 2020.

AURI, founded 35 years ago, focuses on finding new markets for farm products to grow the ag economy through technology and entrepreneurship.

“We’re really excited to see how AVA can help us expand the reach of our mission, which has been historically limited by our number of staff members,” Robinson said. “If it’s not future-proofed, at least it’s looking toward the future in terms of how people access information and can trust that information.”