Skip the Seed Round? New MBA course is turning students into founders at the speed of AI

On a recent Thursday afternoon Alex Zekoff, who co-teaches the new AI Entrepreneurship course at UC Berkeley Haas, wanted to gauge the half-way progress of the class. So he posed a question: How many of you are ready to reach out to ask for startup funding?

“I don’t feel like I need the money yet,” said Daniel Humala, MBA 27, founder of startup Courtship. “In terms of building a product, AI can do this. I’m not sure whether we actually need to fund it.”

Humala’s answer underscored one critical point of the class. Startup founders used to race to get seed funding after they vetted an idea. Now, with agentic AI tools—and the human judgment to know which problems are worth solving in the first place—they are building startups faster and more cheaply than ever. “The bottleneck used to be that to build products, you’d have to hire engineers. Now you can build prototypes and test in the market and get a signal before you go raise and build product capital, said Zekoff,  MBA 17, who created the course with Omri Even-Tov, his former professor. The pair met 11 years ago when Zekoff enrolled in one of Even-Tov’s first classes at Haas. 

Zekoff is now a successful entrepreneur who sold his startup, Thoughtful AI, to New Mountain Capital last May, and he mines from his own experiences to teach students. The class Even-Tov and Zekoff designed is an accelerator program, modeled after Y Combinator, a three-month startup program known for training successful tech founders from companies including OpenAI, Stripe, and Airbnb.

“This class gives students the chance to develop the mindset of an entrepreneur in today’s world.” Even-Tov said. “That means grit, curiosity, the ability to discover new ideas, test products, onboard customers, learn how to sell, pivot when needed, and understand how to get things done. Whether students end up building their own companies or working inside an organization, these skills are becoming essential.” 

AI execs as mentors

Getting into the new course was competitive. Of more than 300 applications from across campus, instructors accepted 27 teams, both solo founders and teams of two or three students. A total of 57 full-time MBA and evening & weekend MBA students enrolled. 

The course taps deeply into the school’s Bay Area location at the epicenter of AI. All semester, a roster of executives from Nvidia, Google, OpenAI, Anthropic, and other companies cycled through the class, some keeping office hours. Today’s mentors included Christina Chavez, MBA 19, who builds AI products as head of international growth at Google, and Vishal Gupta, a product manager at YouTube Shorts discovery, who is integrating LLMs and LRMs (large reasoning models) to improve relevance and personalization of user recommendations. 

Even-Tov interviewed guest speaker Ivan Makarov, a partner at Andreessen Horowitz, who has deep experience as a startup CFO. Makarov noted that most startups die because they run out of cash—not because their original idea was weak; that “cash is truth” and founders need to know their runway cold; and that good financial habits matter—from banking to bookkeeping to forecasting.

“We have more than 130 mentors who understand what it’s like to build using AI in this era,” said Devansh Shah, MBA 27, who is the graduate school instructor (GSI) for the course and co-president of the student-run AI Club. “Five to 10 of these operators, investors and leaders problem-solve with our founders during each class,” he said. Many more hold virtual office hours.

Every week, teams are graded on momentum. “We’re providing community, accountability, and resources,” Shah said. “The only metric that matters is learning over time.”

Building a startup faster

The student teams covered a lot of ground in 13 weeks—from identifying a problem solvable by AI to building a viable product to defining a possible exit strategy. By week three, they were learning key AI engineering concepts like model training, deployment using Amazon Web Services (AWS) SageMaker, and Data Version Control software—a free, open-source system for data, machine learning models, and experiments.

The course provided free access through usage credits for Claude Code, Lovable, AWS, Manus AI from Meta, Notion AJ and other tools, the best tools in agentic coding, deployment, research, and project management. “The idea was to ensure students had every tool the top startups are built with,” Shah said.

Throughout the semester, Andrew Madigan and Justin Keller, both MBA 27, worked with mentors Aayush Agarwal, MBA 25, a senior product manager working on the internal AI platform at Uber, on their startup Populous. Populous uses LLMs to replace laborious customer discovery with simulated populations businesses can use to quickly test their products.

While both were excited about their startup, which was designed to solve pain points they’d both experienced at work, Keller said he quickly learned that others with a similar idea were already getting funded. “There are 10 competitors trying to solve this problem, so we had to find a different angle.” he said, noting one of the down sides of AI: lightening speed startup generation.

Agarwal advised Madigan and Keller to keep going, but to narrow their focus to a specific customer type—which is what they did, quickly. Madigan, a former quantitative analytics manager, said they quickly pivoted. 

“We’ve been able to ship things really fast,” said Keller, who had no coding experience before taking the class. “I couldn’t have done so many of the things that I’ve done and I think I have learned so much just by doing, failing, and doing it again.”

Austin Pruitt, MBA 27, a former BCG consultant, signed up for the class to continue working on his edtech startup to help high school students complete senior capstone projects. He teamed up with Abby Stern, and Kureem Nugent, both MBA 27, to continue the work.

Hands-on experience

Nugent, whose background is in user experience (UX) research, said he was interested in learning to build something from the ground up, with a goal of a career pivot to product management. Stern brought her education background as a former fifth-grade teacher.

“This was a chance to get some hands-on experience and mentorship and work on a tangible long-term project beyond the scope of just a typical class project,” she said.

After learning to use AI tools, the students developed strategies for launching AI products and built a 12-month financial projection for their startups. They went on to address pricing, packaging and positioning; build sales and deal-closing skills; develop a roadmap for their AI company, and, finally, outline an exit strategy.

The class culminated with a Demo Day April 23, where teams showcased their prototypes to a panel of Silicon Valley VCs, alumni, and judges. Of 27 Demo Day teams, 20 teams committed to continue working on their ideas and six teams were accepted to accelerators such as Berkeley SkyDeck and Versel.

“I always have high expectations for our students, but I was still surprised by how much progress they made in just 13 weeks,” Even-Tov said. “That is a testament to the new AI world we live in, but even more, it is a testament to the students’ willingness to push beyond boundaries, move fast, and build things that did not exist at the beginning of the semester.”