Harvard’s $699 Startup Bootcamp Adds AI Avatars for Pitch Feedback

Introduction

Harvard Business School is bringing AI avatars into startup education. Its HBS Foundry program is an eight-week entrepreneurship bootcamp priced at $699. Participants still attend live sessions with instructors each week, but AI versions of those instructors handle feedback during practice pitches and simulated board meetings. The avatars were created with technology from HeyGen.

Key points

• The avatars support repeatable practice. Founders can rehearse a pitch, respond to questions, or simulate a board discussion without waiting for the next live class. The real instructors remain part of the program, but the avatars are used for more frequent practice interactions.
• The format evolved after student feedback. Project director Katharina Rings initially imagined the AI component as something closer to a chatbot. After an early trial, participants said they wanted a more guided experience, leading the program toward avatar-based interactions.
• The simulation still feels artificial. New York Times reporter Sarah Kessler tested the system by pitching an AI-generated version of investor Jeff Bussgang on an “Uber for bananas” idea. Neither the real Bussgang nor his digital counterpart was impressed, and Kessler noticed that the virtual version maintained a noticeably frozen smile.
• Acceptance comes with unease. Bussgang described his digital copy as somewhat “creepy,” while also saying that his students like it. The account does not suggest that the avatars have replaced instructors; instead, they operate alongside live teaching.

Why it matters

The significance of the experiment lies less in the novelty of a talking digital likeness than in where it is being deployed: the repetitive practice layer of education. Startup founders need to refine their explanations, face skeptical questions, and test how a business plan sounds in different settings. An AI avatar can provide an accessible and repeatable practice partner, while human instructors continue to lead live instruction and handle judgments that require broader context.

That division also highlights the technology’s limits. An avatar is not automatically equivalent to the person it represents. The usefulness of its feedback depends on the source material, interaction design, and standards used to evaluate a student’s performance. A stiff expression can make the simulation feel less natural, but it may also remind users that they are interacting with a model rather than the instructor themselves.

Educational institutions will therefore need to explain how such feedback should be interpreted. Students could mistake a confident virtual response for an authoritative conclusion, even though the system may not reproduce every nuance of an instructor’s expertise. HBS Foundry’s approach suggests that the next stage of AI in education may focus less on simply answering questions and more on creating structured, role-based learning experiences. Whether AI avatars become a durable part of instructor support will depend on whether they help learners improve—not merely on how convincingly they resemble a real person.

Source: TechCrunch AI