Harvard Business School has built an eight-week startup boot camp where the faculty are AI clones of its own professors, and the founders taking it say they prefer the clones’ constant availability to the real thing.
The program, called Foundry, launched its current version in April and walks aspiring founders through the standard arc of starting a company: idea generation, validation, fundraising, pitching. What sets it apart is the teaching staff. Several HBS faculty members volunteered to be turned into AI instructors trained on their likeness, teaching style, and area of expertise, though Harvard says the clones are not exact replicas and can draw on broader knowledge when a question falls outside a given professor’s specialty.
Senior lecturer Jeff Bussgang, one of the professors cloned for the program, told the New York Times he finds watching his own AI double teach “somewhat creepy,” but that his students love it anyway. That gap between how the creators feel and how the users respond is the more interesting story here. A PhD candidate at Boston University told Fast Company he values having a version of an instructor he can reach anytime, clone or not. A biotech founder in residence at the University of Massachusetts said Foundry’s simulated board and pitch meetings let her rehearse a hard conversation before walking into the real one.
Access runs two ways. Most of Foundry’s users, including 760 people in its summer cohort, reach it through partner programs at their own universities. Individuals can also buy in directly for $699.
The bigger signal is what Foundry solves for the institutions adopting it. Georgia Tech has folded the platform into a 12-week entrepreneurship workshop, and program principal Harold Solomon was blunt about why: a single instructor can run 20 or 30 meetings a week and no more, and a growing program needs a way to serve every student without hiring proportionally. Mentorship at scale has always been the bottleneck in entrepreneurship education. Foundry is a bet that an AI version of a professor, available at 2am and infinitely parallel, closes that gap better than another teaching assistant would.
It also lands differently than most AI cloning experiments. Posthumous AI performances and AI-generated pastors have drawn backlash for replacing something that was supposed to be irreplaceably human. A pitch-practice partner carries no such expectation. Founders don’t need their AI board meeting to be a real relationship. They need it to be available and useful, and reviews suggest it is both.
Watch for two things next: whether other business schools follow Georgia Tech’s lead and license Foundry rather than build their own version, and whether the same model migrates from entrepreneurship coaching into corporate training more broadly, where the scale problem is identical and the stakes for getting the clone’s judgment right are considerably higher.
