Somewhere in your class is a venture that died in March, and everyone is still clapping. The Kill Board is a prompt your students run in any AI tool: five advisors with colliding doctrines who decide whether a venture should DOUBLE DOWN, PIVOT, or KILL, and who refuse to let any verdict leave the room without dated kill criteria attached. Seven to eight seats per run: five fixed, two or three built for the specific venture.
What your student gets from one run: a verdict they argued for (DOUBLE DOWN, PIVOT, or KILL, with conviction percentages), the Kill Criteria Contract (three falsifiable tripwires with dates), a decision memo, a changelog for their venture's story, and a board report with the minority opinion preserved. About an hour, start to finish.
A student is twelve hours a week into a venture that stopped answering back months ago. You can see it. They can almost see it. But the pitch competition clapped, the campus paper clapped, and every structure in their life rewards continuing.
You cannot be the one to say it. You hold the gradebook, so anything you say about the venture arrives as a sentence about their transcript. And a real live-or-die review takes an hour per venture you do not have, because you have thirty students and one of you.
So dead ventures run on borrowed time all semester, and students learn the worst possible lesson: that finished means graded, and abandoned means failed. Nobody teaches them the actual skill, which is deciding.
A board of five, each owning one way the live-or-die decision fails.
Annie Duke, the quitting scientist. Sunk cost counts for nothing and she says so out loud. Kill criteria: a state and a date, set before the sadness.
Eric Ries, the evidence judge. Sorts traction into evidence and story. The one seat that swings both ways: he will rule "the data says persevere" as readily as "this is vanity."
Paul Graham, the runway realist. Default alive or default dead, run in founder-hours and semesters for student ventures. His question: what date does this end on its own?
Jerry Colonna, the mirror. Reads "why I'm really still in" hardest, and will not vote until he knows who wrote it, the founder or the founder's fear.
Seth Godin, the dip advocate. The seat that keeps this from being a kill machine. Some ventures are just in the dip, and he defends them with criteria, not cheerleading.
Plus two seats built per venture: the Walked-Away Founder, who killed one at this stage and lived, and the Customer Who Left, who judges the traction from the buying side.
Round 1: every advisor tears down the student's written Continue Case, section by section, and votes.
Round 2: they attack each other's verdicts by name, with quotes, and take sides on the central fork.
Round 3: committed final verdicts with conviction percentages and one kill criterion each.
The hand-off: the decision memo, the Kill Criteria Contract, the changelog, the board report.
The exit: every KILL names what the student keeps and their next bet. The verdict lands on the venture, never the student.