For students preparing to graduate, AI brings both possibility and uncertainty. It is opening up new ways to build products and start businesses at a time when many recent graduates are struggling to find their footing in the job market. Learning to use these tools thoughtfully, and gaining experience building something people need, can help students prepare for what comes next.
Jiaxin Pei wants to give UT students that opportunity while they are still in the classroom. This semester, the new iSchool faculty member is teaching INF 385T, Building Large Language Model Applications, in which students have 15 weeks to design, prototype and ship a real AI product. A partnership with OpenAI and support from venture and industry sponsors put tools and practical guidance within reach. The course ends with a public demonstration day, where top teams pitch their products to investors. Whether students launch a startup or take their experience into a future job, the goal is to help them turn their ideas into something real.
Pei joined the iSchool this fall after a postdoctoral fellowship at Stanford’s Institute for Human-Centered Artificial Intelligence. We spoke with him about what students are already building, how Austin’s entrepreneurial community is supporting them and what additional partnerships could make possible.
What inspired the class, and what do you want students to achieve in 15 weeks?
We are at a pivotal moment. Frontier AI is making individuals much more productive, while recent graduates face a difficult job market. I believe it’s our mission to help students become AI-native and reach their full potential. UT has amazing students who want to build things, but they also need the right ecosystem to turn their ideas and prototypes into real businesses. After seeing how the Bay Area ecosystem helps Stanford and Berkeley students build startups, I wanted to help create that kind of support here at UT and in Austin.
The goal is for students to develop their products within 15 weeks and launch their own startups. Even if they don’t launch a company, they will leave with the skills to build with AI and succeed in whatever they pursue.
Can you share a student project that illustrates the course’s potential?
One example is ProtoFlow, an AI-assisted circuit board design tool developed by a student in the class. It aims to reduce the manual work involved in finding components and creating circuit designs while keeping engineers in control. It has already attracted real users and generated revenue, and we’re helping the team grow it into a successful startup. There are many other exciting projects, which we look forward to showcasing at our demo day at the end of the course.
How are partners helping students bring their ideas to life?
Through our partnership with OpenAI, students receive Codex access and API credits to build and ship their products. Support from Antler, Standard VC and HydraDB, alongside UT’s Longhorn Venture, Texas Innovation Center and AI Studio team, helps create a community around that work. Guest speakers and UT alumni also bring experience students can draw on as they develop their ideas.
Those connections help students think through questions that come up when building a business: Who needs this product? What should we build first? How do we know it’s useful? Access to tools gives students room to experiment, while conversations with founders and investors help them decide where to take their projects next.
What happens when students work with industry?
Students bring fresh ideas, and people working in industry bring a close understanding of the problems their organizations face. Many teams are developing products to help businesses adopt AI, so those conversations can help students understand where their work could be useful and what it would take for someone to use it.
For companies, it’s a chance to get to know how students think and build, while exploring ideas together. For students, it makes the work more meaningful when they can connect what they’re learning to a problem someone actually needs solved.
What could help promising projects grow beyond the course?
The end of the semester doesn’t have to be the end of a promising idea. Continued access to AI tools and computing resources, along with small project grants, could give students time to keep testing and improving what they’ve built.
Regular guidance from mentors and opportunities to test products in real settings would help teams learn from users and decide what comes next. That experience matters whether a project becomes a startup or helps a student find the direction they want to pursue.
To learn more about the course and the projects taking shape, contact Pei at jiaxinpei@utexas.edu.