Define · Create · Analysis & Finance

MVP and the test that proves it

The smallest feature set that tests your riskiest assumption (built for your first few users), an explicit cut list, and a single pass/fail test card with a numeric threshold.

You receive: A JSON object { problem, riskiest_assumption, minimum_feature_set[], cut_list[], feature_rationale[], fidelity, test_card{hypothesis,simplest_test,pass_threshold} } checked so the kept and cut sets are disjoint, every kept feature is justified, and the test carries a number.

Part of Validate Demand

Opens soon

Cost20 credits

AcceptanceAutomated check against your inputs

ProtectionHeld until verified delivery

This objective is verified and ready. It opens soon, once sign-in and payments are live.

Deliverable interface

The exact vocabulary the automated check enforces: keys, tokens, entry points, and worked examples. Generated from the verification source.

{
 "buyerParamKeys": [],
 "checkNames": [
  "parses_json",
  "problem_present",
  "riskiest_assumption_present",
  "minimum_feature_set_min",
  "cut_list_present",
  "mvp_cut_disjoint",
  "feature_rationale_covers_mvp",
  "fidelity_valid",
  "test_card_complete",
  "pass_threshold_quantified",
  "no_em_dash"
 ],
 "constants": {
  "_YC_EM_DASHES": [
   "—",
   "–"
  ]
 },
 "deliverableLabel": "yc_mvp_and_test",
 "documentKeys": [
  "cut_list",
  "feature",
  "feature_rationale",
  "hypothesis",
  "minimum_feature_set",
  "pass_threshold",
  "problem_link",
  "simplest_test",
  "test_card"
 ],
 "enumLiterals": [
  [
   "low",
   "high"
  ]
 ],
 "submission": "a JSON document (submitted as a string), validated as data; every check below must pass",
 "tier": "doc-validator"
}

Example

A sample of what this objective produces. Your result is generated for your inputs.

Problem

Founders cannot tell whether an AI agent actually did a task correctly before they pay for it.

Riskiest assumption

Founders will trust a machine-checked proof enough to accept and pay without manually reviewing the work.

Minimum feature set

Post one narrowly scoped task type (structured data extraction from a document)
Return the deliverable bundled with an automated pass/fail verification artifact
Buyer reviews the proof and clicks accept and pay

Cut list

Multi-category support: only one task type at launch
Agent ratings, reputation, and self-serve matching
Automated payments, escrow, and payouts
Dispute resolution, SLAs, and refund tooling

Feature rationale

FeaturePost one narrowly scoped task type

Problem linkkeeps the test focused on the trust question, not breadth

FeatureReturn the deliverable with a pass/fail verification artifact

Problem linkthe artifact is the proof of correctness buyers cannot get today

FeatureBuyer reviews the proof and clicks accept and pay

Problem linkcaptures the accept and pay decision that tests the riskiest assumption

Fidelity

low

Test card

HypothesisBuyers accept and pay after seeing the verification artifact

Insight soughtwhether verifiable proof, not just the raw output, drives the accept decision

Simplest testrun 20 real paid operations of one task type through a manual kernel and measure accept and pay

Pass thresholdat least 60 percent of the first 20 delivered operations accepted and paid

Mediumconcierge run with real buyers

Get early access to STUD the day it goes live.