Specialty centers · AI Venture Build
Free briefs for a real company shape with a real unit economic, that happens to use a model. Students cost one unit of value at three volumes and find where gross margin breaks, and they write down what would kill the idea before they run the first customer interview. VEN-04 and VEN-01 have single-brief PDFs; every brief is in the bank PDF; the other ten are full on the page. No account. No student data leaves your school.
A venture study, not a company
On a school assignment students take no money, sign nothing, register no entity, hold no customer data, and ship to no one who might rely on it. A letter of intent is not revenue. Customer interviews are introduced honestly and arranged through the school. HQWBL paperwork stays school-owned — the brief is the student's project, never the compliance record.
How to run one
Worked sample · printable PDF
Your venture delivers some unit of value — a summarized document, a matched candidate, a graded assignment, a triaged ticket. What does one of those actually cost you to produce, all in, at three different volumes?
Featured · printable PDF
Write down what you believe about your customer's problem, and what you would have to hear to stop believing it. Then go find out. After a minimum set of interviews, report what you were wrong about — and whether the idea survived, changed, or should be abandoned. A student who finishes this brief with the same idea they started with has either found a real problem or failed to listen. The evidence has to show which.
Name the one customer you would serve first, and defend why the larger market comes later — or never. A wedge is a decision about who you are willing to disappoint.
Design the version of your product that uses no machine learning at all — a form, a spreadsheet, a rules engine, an off-the-shelf tool, a person doing the work. Then say what the model has to beat, on a named metric, to earn its place. If it does not beat the simple version, that is your answer.
What do buyers pay today for the thing you would replace, what did they tell you about budget, and what happens to your margin at that price given what VEN-04 found?
Then mark the parts you do not have, cannot get, or have no permission to collect. Is what remains actually a moat, or is it a liability with a storage bill?
Your product will be wrong sometimes. Design the disclosure, the human review step, and the failure experience that make a buyer say yes anyway — then test whether the design actually works on someone who has not seen it before.
List the rules that could stop your launch — in Virginia schools, in a regulated industry, or in a consumer market — and say which one you would hit first. Not a survey of AI policy: the specific requirements that apply to this venture, this buyer, this year.
Deliver a five-minute pitch that includes an explicit risk slide, and survive the questions. The test is not whether judges liked it — it is whether a skeptical listener could find a hole you had not already named.
Where does the capstone already satisfy a scored criterion, where does it need reshaping, and where would the student be scored on something they have not done at all?
Assemble the artifacts a work-based learning coordinator would recognize for an Entrepreneurship experience — and be precise about which pieces are the student's to produce and which belong to the school.
Then apply them honestly to what you actually have. Would your own plan survive its own test?
Six-dimension rubric
Scope unclear? Numbers off? Tell us here — you do not have to leave an email.