I spend a lot of time talking to founders who are trying to solve problems that don't actually exist. They build massive, compute-heavy models for markets they don't understand, hoping the hype will carry them to a Series A. But if you want to see what actual innovation looks like, you have to look away from the tech hubs and toward places like Sterling, Colorado. That is where Xue “Alice” Dong is doing the hard work of bridging the gap between high-level AI research and the dirt-under-the-nails reality of rural farming.
Dong is an extension regional specialist for Colorado State University. In plain English, she is the person responsible for making sure university breakthroughs actually help people who grow our food. Her territory covers ten counties in the northeast corner of Colorado—an area where the average farmer is more likely to be worried about cattle prices or soil health than the latest LLM benchmarks. But Dong realized early on that these farmers were being left behind by the digital revolution. Their kids are moving to cities, their technical support is non-existent, and the AI boom felt like something happening on another planet.
The Pragmatic Build: AgriTourismAI
Instead of just giving a speech about how AI is the future, Dong built something tangible. She developed the AI-Powered Agritourism Chatbot (AgriTourismAI.com). For those unfamiliar with the term, agritourism is when a farm opens up to the public for things like corn mazes, pumpkin patches, or educational tours. It is a high-margin way for a family farm to stay afloat when crop prices dip, but it requires a massive amount of business planning, marketing, and legal navigation that most ranchers aren't trained for.
What I love about Dong’s approach is her technical restraint. She didn’t build a flashy mobile app that requires a 5G connection and the latest iPhone. She built a web-based tool in Python because she knew her users might be on older desktop computers or tablets. Under the hood, she used Retrieval-Augmented Generation (RAG). This is a founder’s best friend. Instead of letting the AI hallucinate advice about liability insurance, the RAG system forces the bot to look up information from a curated set of trusted CSU Extension documents first. It is AI with a leash, which is exactly what a business owner needs when they are asking about permits and risk management.
Solving the Digital Divide
We talk a lot about the “digital divide” in tech circles, but we rarely see people actually crossing it. Dong lives in the community she serves. She isn't commuting from a glass office in Denver; she is at the county fairs and the farmers' markets. She has run workshops for over 200 entrepreneurs across 32 counties, teaching them how to use tools like Claude and Gemini to handle the administrative overhead that keeps them out in the fields until 9:00 PM.
Her results speak for themselves: 90 percent of her workshop participants reported trying at least one new AI task in their business. That is a higher conversion rate than most SaaS onboarding flows. Why? Because she isn't selling them “transformation”—she is giving them a tool that helps them write a business plan for a corn maze or translate complex weather data into plain language.
Building for the "Non-User"
One of the most impressive parts of Dong’s work is her collaboration with the National Center for Atmospheric Research on Project Pythia. She helped build an AI assistant that takes raw, technical airport weather reports and turns them into something a farmer can actually use to make planting decisions. It even tells the user when data is missing rather than making a guess. In an industry where one bad call on a frost date can cost a family their livelihood, that kind of honesty in an AI system is mandatory.
For builders, there is a massive takeaway here: the most underserved markets are often the ones with the most to gain. We have enough AI writing assistants for marketing agencies. We don't have enough tools helping the people who manage our natural resources and food supply. Dong has secured over $250,000 in grants to prove that this work matters, and she has been recognized globally by the UN and IEEE for her AgriPivot AI Agent.
The Founder Perspective
If you are a founder looking for a moat, look at what Alice Dong is doing. She hasn't just built a wrapper; she has built a localized, context-aware system backed by decades of institutional research and personal relationships. You can't disrupt that with a bigger GPU cluster. You compete with that by being on the ground, understanding the specific pain points of a niche community, and building the simplest possible solution that solves the problem.
“We act as a bridge,” Dong says. “We bridge our campus to the rural communities we serve.”
That bridge is exactly what the AI industry is missing right now. We have the researchers (the campus) and we have the real-world problems (the rural communities), but we have very few people willing to live in the middle and do the translation work. Dong’s framework for “Governance-Aware Design,” published in 2026, should be required reading for anyone trying to deploy generative AI in small business environments. It isn't just about whether the technology works; it's about whether it can be used responsibly by someone who doesn't have a computer science degree.
What Builders Should Take Away
- RAG is the standard: If you are building for high-stakes industries like agriculture or law, your AI must be grounded in a verified knowledge base. Hallucinations are a deal-breaker.
- Accessibility matters: Don't build for the latest hardware. Build for the hardware your customer actually owns.
- Local context is the moat: Data is everywhere, but the ability to translate that data into a specific local outcome (like a corn maze permit in Logan County) is where the value lies.
- Trust is the currency: Dong’s success comes from her role as a trusted extension specialist. In AI, the technology is often less important than the trust the user has in the person showing it to them.
Alice Dong is proving that the next wave of AI isn't just going to happen in Silicon Valley labs. It is going to happen on buffalo ranches and in small-town community centers. The question for the rest of us is whether we are willing to step out of the hype cycle and build something that actually helps someone grow a crop or start a business.
Read the original at IEEE Spectrum →