“AI has allowed us to develop the product a lot faster and has allowed us to actually solve the problem that people are facing in a more efficient manner,” Kavula added. “I think, personally, it’s going to help a lot more businesses come from just ideas … and help people build more things that customers actually want, which I think is the biggest goal at the end of the day.”

Lukas Garcia: Clemson University

Garcia described himself as the son of Argentine immigrants who emigrated to the U.S. when their homeland took an economic nosedive in 2001. His father is now a professor in Clemson’s Department of Chemistry in South Carolina, where Garcia is gaining attention for Crystal-XG, his award-winning early-stage venture. Garcia said he’s grateful to America250 for shining the spotlight on innovation and to live in a country full of “brilliant minds, brilliant companies, scientists, and engineers.” 

“I think the United States is the best country in the world,” he said. “It’s completely understated that the opportunities that we have in the United States, and so therefore, I think, like the American250 campaign is fantastic, and recognizing the fact that, you know, we do live in a country, and we are very fortunate, as U.S. citizens, to have all of these opportunities around us. And I don’t think that we should take that for granted.”

Garcia said he had been doing research for several years prior to attending Clemson, and grew frustrated one summer as he worked in a lab on crystal synthesis. By summer’s end, he said he had only around 10 publishable crystal results to show for 500 experiments conducted. That’s when Garcia got the idea that “if some of the best-trained humans don’t know what they’re doing, then a computer or a machine learning algorithm probably can.”

“I realized it was a really inefficient process. And not only that, I realized that there were vast applications in pharmaceuticals, material science, energy, catalysis,” he said. 

To solve the inefficiencies, Garcia said he created an AI algorithm that can output results in under a minute, compared to the six months it can take without the model for a candidate molecule to become fully crystallized. There are now two U.S. labs and 11 worldwide running predictions with Garcia’s algorithm.

“A scientist inputs the molecule that they want to crystallize. And so our model converts that molecule into chemical language that it can understand, and from there, it’ll pretty much find the synthetic pathway necessary to crystallize it,” he said. “So it visualizes what that molecule looks like in 3-space, has a few different thermodynamic and chemical properties that we calculate, and then from there, it’s just finding the best pathway to crystallize it.”

Garcia emphasized hopes that by cutting costs and inefficiencies, his AI algorithm can help revitalize pharmaceutical manufacturing in the U.S., an industry he suggested has been hit hard by federal funding cuts. 

“We are super interested in helping out some of these research labs here within the United States,” he said. “Pharmaceutical manufacturing — we really want to bring it back to the United States, you know, chemical breakthroughs, scientific breakthroughs, and whatnot, bringing novel therapeutics to patients, that’s the main thing I want to drive home.” 

Sanjana Kavula and Adhira Tippur: Rice University

Sanjana Kavula and Adhira Tippur are from Rice University in Texas. They worked as a team to develop an AI-powered patient intake and front-desk automation platform called Kairos that is designed for dental clinics. Tippur credited the U.S.’s “powerful” infrastructure for driving both her startup’s success and innovation more broadly in the country. 

“The U.S. is uniquely positioned when it comes to innovation, especially as a student founder like Sanjana and myself — it’s really something here where an idea starts from a very personal problem, and it can become a real company, because the infrastructure essentially exists to really support that,” the student said during an interview. 

“There are a lot of university startup competitions, mentors that we’ve had access to, early customers willing to pilot with us, investors willing to take bets on young founders. That ecosystem in general, that America provides, is quite incredibly powerful, especially to us,” she added. 

Tippur said she initially became aware of the administrative burden on healthcare workers growing up, when working in her mother’s dental office in the Rio Grande Valley. As she grew older and started working more in healthcare technology, Tippur said she kept coming back to that direct experience and realizing it “really impacted” patients’ access to quality care. 

“I remember feeling like everyone was doing 10 things at once, and so the front desk especially always felt quite overwhelmed,” she reflected. “Phones are really nonstop, lots of patients just trying to schedule in. A lot of insurance coming in, people showing up in person, looking for help, and just general emergencies that would keep popping up.”

Tippur joined forces with Kavula to create a system to solve the problem. Kairos works by plugging an AI program into clinics’ existing practice management software, such as Open Dental, Dentrix, or Epic systems, Kavula said. The two students said they customize Kairos to each clinic’s system to adapt to varying workflows. Eight clinics are now using Kairos, with more set to adopt the technology in the coming months, according to the pair, whose operations are based largely in Texas. 

The students said clinics use Kairos to schedule appointments, process calls, and manage insurance workflows, among other tasks. 

“With Kairos, what we do is we’re an end-to-end patient intake system for healthcare clinics,” Kavula said.

Leo Guan and Ayush Banerjee: Georgia Institute of Technology

Guan and Banerjee worked as a team to develop Fractal Semi after realizing the primary issue with semiconductor innovation was that American manufacturing had failed to keep up with chip design, due to an overreliance on offshore producers. 

Banerjee explained that advanced semiconductor chips are created through complicated recipes with hundreds of steps. Much of that knowledge is currently “stuck” offshore, he said, and stored in facilities outside of China, creating a national security supply chain risk as U.S. semiconductor foundries lag behind. The students cited Taiwan Semiconductor Manufacturing Company as the U.S.’s main advanced semiconductor supplier, meaning cutting-edge AI and defense chips have been largely outsourced to overseas manufacturing. 

To close the supply chain gap and spur domestic manufacturing, Guan and Banerjee are working on an AI model to simulate the process. By replicating the foundry virtually, they hope to radically simplify the process, slashing costs for U.S. manufacturers. The $25,000 award from American250 will help them understand one step in the semiconductor process, which they say will allow them to extrapolate. 

An integral piece of Fractal Semi includes plans to embed sensors in existing semiconductor manufacturing machines inside foundries. Doing so will allow the researchers to collect data on how the semiconductor chips are built, allowing them to train the AI model. The pair hopes it will cut down the costly and lengthy trial-and-error processes engineers often grapple with when working to develop cutting-edge chips. 

“By having an AI model that knows the physics of how to create a semiconductor, it can basically help you gain the knowledge of how to create one without doing trial and error that can be super expensive,” Guan said, adding that trial and error “can take five years and billions of dollars.”

The students emphasized the revolutionary role AI has had on technology, both through accelerating and democratizing innovation. And they hailed the U.S. as a nation that provides unique opportunities for entrepreneurs, praising Palantir’s tech fellowship, Nvidia’s inception program, and even the platform OpenAI provides, for “giving back.”

THE AGE OF AI IS HERE — HOW SHOULD IT BE REGULATED? 

“There’s a lot of opportunities to innovate, to create our own specific AIs, using the resources that these great companies have provided,” Banerjee noted. “I don’t think that’s possible in other countries around the world, at least for now.”

“AI is super important, because essentially, if you train an AI well enough, it sort of understands reality, right?” Guan added. “So essentially, I think that puts you at a point where you don’t really have to run live experiments anymore, you can have these things called digital twins that kind of emulate what happens in reality. … Discovering new things just becomes a lot easier with AI and a lot more inexpensive.”