A prestigious business school is now letting students practice their startup pitches in front of a computer-generated double of a real instructor. It is an experiment in using technology to scale personal feedback, giving learners a way to rehearse high-pressure conversations whenever they want, rather than waiting for a professor’s limited office hours.
Harvard Business School has launched an eight-week online entrepreneurship program that uses digital avatars to critique student work. These avatars are created by a company called HeyGen, which uses software to produce a realistic video of a person that speaks and reacts to input. While the program includes live sessions with human instructors, the avatars handle the heavy lifting during practice pitches and mock board meetings. Students present their business ideas to these digital versions of real industry experts, who then offer feedback based on their training. Early reports indicate that participants generally find the tool helpful for practicing their delivery, despite the fact that the digital replicas can sometimes look a bit stiff or unnatural during the conversation.
How digital instructors learn to critique
To build these avatars, the developers use a sophisticated piece of software known as a model—a digital engine that has been taught to recognize patterns in human language and behavior by analyzing massive amounts of examples. First, the human instructor provides a large amount of data, including their own voice and video footage, allowing the software to create a digital puppet that mimics their appearance and mannerisms. The underlying intelligence behind the avatar is trained on specific materials and expert knowledge, enabling it to recognize the context of a student's pitch. When a student speaks, the system converts that audio into text, processes it through the trained logic to determine if the pitch makes sense, and then generates a spoken response. The visual avatar then animates its mouth and facial expressions to match that response, creating the illusion of a face-to-face coaching session.
Using AI for instruction solves a classic bottleneck in education: one teacher cannot watch a hundred students practice simultaneously. By using an avatar, a student can fail, refine, and try again as many times as needed in a private, low-stakes environment. This marks a shift toward on-demand mentorship, where the goal is not to replace human experts, but to ensure that everyone gets as much practice as they need. The big question moving forward is whether these digital proxies will become a standard way to build confidence before the real-world high-stakes meetings, or if they will always feel like a pale imitation of a human teacher's genuine insight.
Liked this one? The next lands at breakfast.
Every story in tomorrow's AI news, rebuilt in plain English — five minutes, sources linked, free forever.
By joining you agree to receive Article's daily newsletter — unsubscribe in one click. Privacy