What does the buyer agree before work starts?
Buyer-facing language stays simple: one offer, one order, and the result with proof.
- Inputs
- The minimum material the buyer must provide.
- Full price
- The total price for the agreed scope, shown before commitment.
- Literal result
- The exact deliverable the buyer receives.
- Deadline
- When the result and its evidence are due.
- Acceptance
- The test or authorized reviewer empowered to accept.
- Remedy
- Agreed rework, refund or handoff if acceptance fails.
The family-book and restaurant examples use fictional terms to make this structure tangible. They are not offers for sale.
What counts as done?
A model response, generated file or sent deliverable is not completion. Only Verified complete is success.
Every job needs one accountable owner, bounded operators, available helpers, an authorized verifier and a named escalation path. Failed checks remain unresolved until rework, escalation or the agreed remedy.
What are the three records behind the order?
A versioned Method Spec defines the reusable service. A Work Order records the job-specific scope, owner, decisions, state and exceptions. A Completion Packet contains the deliverable, acceptance result, evidence, limitations, rework history and signoff. These are planned internal records behind one simple purchase.
How does the adaptive interview work?
It seeks the smallest valuable repeatable service inside an expert practice. After each answer or artifact, the next question should reduce material commercial or execution uncertainty.
The design uses 17 high-value core questions with room for up to four targeted follow-ups or evidence requests. Twenty-one is an interaction budget and metaphor, not a rigid form, validated science or guarantee. The interview may stop early.
What do experts contribute and control?
Experts provide examples, judgment boundaries, failure modes, corrections and the acceptance rubric. They judge outputs rather than build node graphs.
Who owns and prices a method?
The design intent is for experts to retain their method and agree scope, price, turnaround, review level, attribution, corrections, use and payout terms before a pilot. Those commercial and ownership policies are not finalized. Private import for existing services is planned, not built.
What should improve with each accepted job?
Corrections, exceptions, rework, cost and acceptance history should improve the next Method Spec version and reduce expert effort per accepted result.
Why could the economics improve?
Better models may lower delivery cost without changing the promised result. That is a hypothesis to measure through expert minutes, rework, acceptance and gross profit per job. Expert decisions, evaluation cases and completion history—not model access alone—are the intended durable asset.
How would Methora earn?
The intended model aligns revenue with accepted work rather than seats or tokens. Actual prices, fees, payout timing, refunds, cancellation and out-of-scope review policies must be agreed before a pilot. The website sets no live commercial policy.
What should someone share today?
Only a short, non-confidential description in the interest form. Do not submit customer documents, credentials or work-order inputs.
What would a pilot require?
Input permissions, model use, access, retention and handling terms must be agreed before any material is shared. The privacy policy covers website signup, not future customer-document processing.
How does one method become a market?
The long-term ambition is a trusted market of proven methods callable by people, service firms, software and AI through the same work contract.
How is wider access earned?
First a controlled pilot and repeated verified completion; then a firm’s catalog of proven methods. Later, methods could compose under one total price and parent remedy while each output keeps its own check and attribution. Public discovery, software invocation and a broader marketplace are vision, not current capabilities.
Who can I contact?
Methora is built and run by two individuals, Dezső Mező and Miklós Huber. For cohort or method questions, contact miki@launchloop.xyz.