A founder once spent thirty minutes walking a friend who mentors startups through a business plan. When the pitch was done, the mentor said: "The flow was natural, and the opening was strong. You might just want to tighten up the numbers a bit." The founder jotted down a few notes and left the room. One practice pitch, complete.
In an actual investor meeting, the questions start before the second slide even loads: "If your repeat-purchase rate is at that level, how many months does it take to recoup customer acquisition cost?"
The mentor's feedback wasn't lazy. It was exactly what that setting could produce. There's a different category of feedback that setting simply can't deliver.
The quality of feedback is set by the feedback-giver's stake in the outcome
When an investor questions a founder mid-pitch, the question isn't "was this presentation well put together." It's "how many ways can this business lose me money." When a board member reviews a strategy, the first question isn't "is this logically sound" — it's "if this fails, how do we explain it."
A friend, a mentor, an early-stage investor, and a board member can hear the exact same pitch and interrogate it from entirely different angles. Some of that gap comes from differing expertise, but the more decisive factor is what that person is actually on the hook for. Someone with their own money or legal liability on the line has no reason to be generous. So they ask differently.
Practice pitches land softer than the real thing because the person giving feedback isn't a stakeholder in the business. Without skin in the game, there's no reason to ask the hard question.
What Harvard packed into $699
This past August, Harvard Business School launched a startup bootcamp called "HBS Foundry" priced at $699. The program's AI avatar is modeled on an HBS instructor and gives feedback on students' practice pitches and board-simulation exercises.
Compared to the cost of sitting in an actual HBS classroom, $699 is a modest number. Look more closely at what's actually inside that price tag, and it's the experience of being questioned, HBS case-method style, about the assumptions underneath your business.
In the case method, an instructor listens to a student's argument and then hunts for the scenario where the underlying assumptions collapse: "What happens if this market turns out smaller than expected?" "Why would a customer bother absorbing the switching cost?" If the avatar genuinely replicates that approach, founders get to practice fielding objections they almost never hear from a friendly mentor.
Plenty of courses teach you how to analyze a strategy. Very few put you through the drill of defending that strategy against a real stakeholder's objections. The $699 program is an attempt to fill that particular gap.
Can the avatar actually ask hard questions?
There's one telling clue.
"What's your repeat-purchase rate?" and "If your repeat-purchase rate sits at that level, what's the actual case that this business is sustainable?" are entirely different questions. The first just confirms a number. The second challenges the business model itself. Most AI feedback tools tend to default to a "here's what worked, here's what to improve" format — because designing for a response that makes the user feel comfortable is the default instinct behind most of these products.
How far the HBS avatar has actually strayed from that default is hard to verify from the outside. Results here depend heavily on training data and prompt design, and not much of that has been made public. The reasonable move is to wait for real user reviews to accumulate before spending the $699.
Why Korean founders' feedback skews friendly
This same gap exists in Korea, too.
Trace where early-stage founders and solo entrepreneurs actually get feedback on their business plans, and it mostly funnels through three channels: personal networks, mentoring at government-run startup support centers, and accelerator screenings. Personal networks default to encouragement — there's simply no incentive to say something discouraging. Startup-center mentoring tends to focus more on polishing the pitch deck or paperwork than stress-testing the business itself. Accelerator screenings are sharper, but founders only really hear that feedback after they've been rejected — by which point the opportunity has already passed.
There's almost no channel, at the stage when founders are actually preparing, that delivers feedback from someone who thinks like a stakeholder.
You don't need an AI avatar to partially close this gap. Changing how you ask for feedback works too. Instead of "take a look at my pitch," try "find me three reasons not to invest in this business" — the same friend will suddenly interrogate it from a different angle. Assigning a role — "look at this the way an early-stage investor would," "if you were a competitor, how would you undercut this" — repositions where the feedback-giver is sitting.
Even more useful is tracking down an actual stakeholder. A former professional from the relevant industry, a demanding potential customer, a founder who's already failed in a similar market — none of them have any reason to be gentle. Their objections belong to a different category than a friendly mentor's.
What HBS Foundry is selling for $699 is a simulation of that stakeholder's seat. Whether AI can convincingly play that role is something that will get tested over time. What's already clear is this: the mere existence of a product like this says something about just how rare it is to get a genuinely hard question in return.
