University prestige has long been sold as a rare experience: a campus, a room, a professor, a handful of selected students, and the network that remains after the degree. Harvard Business School is now testing a stranger and more scalable version of that promise. In its HBS Foundry program, founders can practice in front of AI-generated avatars that resemble real instructors. These are not professors being replaced by machines in a lecture hall. They are digital doubles used to rehearse the most anxious moment in entrepreneurial life: persuasion.
TechCrunch reported on August 22, 2026, that the Foundry bootcamp, an eight-week online program priced at $699, includes avatars built with HeyGen technology. Weekly sessions remain led by humans, but the avatars provide feedback during practice pitches, simulated board meetings and preparation exercises. The detail matters: Harvard is not simply selling a course video. Harvard is selling repetitions, and repetitions are exactly what entrepreneurs pay dearly for when they come from an investor, mentor or recognized professor.
The Real Product Is Repetition
In traditional education, personalized feedback is the scarcest resource. A course can be recorded, a case can be reread, a method can be summarized. But watching a founder hesitate, drift, defend a fragile assumption or miss a market slide takes human time. That is the time that drives improvement. A pitch is not only a document. It is a performance under pressure, with rhythm, presence, answers to objections and the ability not to lose composure.
Foundry’s avatars fit into that logic. According to TechCrunch, a New York Times reporter tested the experience by pitching a deliberately absurd idea, “Uber for bananas,” to an AI copy of Jeff Bussgang, co-founder of Flybridge Capital and senior lecturer at Harvard. Both the real Bussgang and his digital double were reportedly unconvinced. The reporter noticed the avatar’s notably frozen smile, and Bussgang himself admitted that watching his double was “a little creepy,” while adding that his students loved it.
That paradox tells the whole story. The avatar does not need to be perfectly human to be useful. It needs to be credible enough to create tension, available enough to allow practice, and consistent enough to repeat a method. In a business school, that changes the marginal cost of feedback. A professor cannot listen to a thousand pitches in a row. His double can. The question is therefore not only pedagogical. It is industrial.
Harvard Turns Its Method Into a Platform
The official HBS Foundry site describes the program as an “AI-native” digital workspace designed to help founders move from raw idea to fundable pitch. The promise combines business model development, market validation, pitch deck building, live sessions, community and AI tools. The enrollment fee listed in the bootcamp application is $699 if the candidate is accepted. At Harvard scale, that ticket looks almost accessible. At the scale of the entrepreneurship training market, it is highly strategic.
Harvard does not appear here as a traditional university jealously protecting its scarcity. It appears as an educational brand building a software layer around its know-how. The name, methods, pitch scenarios, evaluation standards and faculty figures become distributable components. It is the pedagogical equivalent of a studio turning its characters into interactive experiences: the value is no longer only in the live event, but in the ability to deploy a branded presence wherever the user practices.
Poets&Quants, which covered the news in its MBA roundup, underlined this shift: HBS Foundry targets founders who want to move from idea to stronger pitch, with an opportunity to compete for $100,000 in funding. The program does not replace an MBA and does not claim to. It creates a different category: a short, less expensive, heavily branded, AI-augmented training experience that promises access to part of the language and rituals of the Harvard ecosystem.
The Professor’s Face Becomes an Asset
The most delicate dimension concerns likeness rights and trust. A pedagogical avatar is not simply a chatbot with a voice. It is the reproduction of a person, their gestures, their authority and their symbolic capital. In film, music and advertising, the question of digital doubles is already explosive: who authorizes the reproduction? For how long? In which contexts? With what compensation? With what right of withdrawal? Education will have to answer the same questions.
The Harvard case is less brutal because the instructors appear to participate voluntarily and the avatars are confined to a precise pedagogical setting. But the precedent matters. If a school can create a version of a professor available 24 hours a day, that professor also becomes infrastructure. Their method can be replicated, localized, integrated into new products and perhaps one day separated from their real schedule. The contract between institution and instructor will have to become much more explicit.
For students, the question is different: does the avatar truly improve learning, or does it merely create the feeling of premium access? The early reactions reported by TechCrunch are favorable, in part because participants wanted a more guided experience than a simple chatbot. That makes sense. A founder does not only need information. They need an interlocutor, even an imperfect one, to feel the friction of a difficult meeting.
The Risk of the Licensed University
The model can nevertheless slide. If prestigious universities turn their expertise into avatars and microprograms, they can democratize a portion of access. That is the bright promise: a founder who cannot afford or access an MBA can test a method, receive feedback and learn the vocabulary of investors. But universities can also multiply peripheral products where the logo’s prestige carries more weight than the depth of support.
The boundary will lie in the quality of the pedagogical design. A useful avatar is not a video gimmick. It must ask good questions, spot weaknesses, protect confidentiality, avoid hallucinations, avoid overpromising and direct users to humans when the stakes require it. HBS Foundry says it protects founders’ ideas and confidentiality, and its official pages emphasize security and intellectual property. Those guarantees will be essential if entrepreneurs are expected to entrust concepts, data and strategies to an AI platform.
The edtech market is already watching. If Harvard can make high-level feedback more repeatable, other schools will follow. Accelerators, incubators and corporate programs will have to explain why their mentoring hours cost more if part of the practice can be automated. Conversely, the best human mentors may become even more valuable: no longer for repeating the same critique ten times, but for stepping in when the avatar reaches its limit.
A More Accessible School, or a More Commercialized One?
The Foundry experiment condenses the great ambiguity of AI in education. It can open access to guided practice that many people would never have obtained. It can also turn teachers into licenses, students into training data and universities into software brands. Everything depends on transparency: what the avatar knows, what it does not know, what it records, what the instructor accepted, and what the student is really buying.
Harvard chose a clever terrain: the startup pitch. It is a field where repetition is useful, feedback can be structured and the emotional relationship with the professor-avatar remains less intimate than in therapy, a children’s classroom or a literature course. But the precedent is here. After online courses, certificates and platforms, the university is entering the age of the double. The future of education may not be a campus without professors. It may be a campus where the professor also exists in a repeatable, calibrated, commercializable version. The question is whether that copy will help the student grow, or only the platform.
