Starting

What actually has to be true before a company exists?

1. 01Aurygine synthesisEvidence quality: high

   Customer-funded growth: making the buyer your investor

   Aurygine's own synthesis across observed practice. No single work is reproduced.

   Extracted pattern

   • A customer who pays before the thing exists has told you more than any survey could.
   • Deposits, pilots and prepaid pilots are financing instruments, not just revenue.
   • If nobody will pay in advance, the problem is usually urgency, not price.

   Decision pattern

   • Situation. Founder needs money to build, has no capital and no track record.
     Move. Sell a paid pilot to three buyers with a fixed scope and a delivery date, and build only what the pilot requires.
     Trade-off. The first version bends towards three specific buyers and may not generalise.
     Breaks when. The product needs heavy upfront infrastructure before any value exists, or the buyer cannot legally prepay.

   Practical implication

   Before raising or spending, test whether the problem is urgent enough to be prepaid. Prepayment is the cheapest validation instrument available to a founder and it funds the build at the same time.

   Counter-pattern / contradiction

   • Capital-intensive and deep-tech companies routinely cannot be customer-funded early; milestone financing fits better.

   Failure modes

   • Building for six months on savings, then discovering the urgency was imagined.
   • Discounting the pilot so heavily it proves willingness to try, not willingness to pay.

   What was measured

   prepaid pilots signed · cash collected before build · pilot-to-contract conversion

   Stage relevance

   Idea · Pre-revenue · First customers

   Source

   Aurygine · 2026-01-12
2. 02SourcedEvidence quality: high

   Manual first: why early unscalable work is the strategy

   Summarised in Aurygine's words from a publicly readable source, linked below.

   Extracted pattern

   • Recruiting the first users by hand is the work, not a workaround for the work.
   • Manual delivery is a research instrument: it shows you what to automate and what to drop.
   • Founders who refuse the unglamorous phase usually never learn what the product should be.

   Decision pattern

   • Situation. A new product has no users and no distribution.
     Move. Recruit the first users individually, deliver the outcome by hand, and observe every failure directly.
     Trade-off. It does not scale, consumes the founder's time, and looks inefficient from the outside.
     Breaks when. Manual work is masking a product that fundamentally does not work, or when it continues long after the pattern is known.

   Practical implication

   Treat the manual phase as paid research. Serve a handful of users personally until the repeatable part of the work is obvious — then automate exactly that part and nothing more.

   Counter-pattern / contradiction

   • Regulated or safety-critical products cannot be delivered by improvisation; the manual phase has to be designed, not improvised.

   Failure modes

   • Waiting for a launch moment instead of finding ten users this week.
   • Automating a process before you have run it by hand often enough to know its shape.

   What was measured

   hand-recruited users · time to first delivered outcome · retention of hand-served cohort

   Stage relevance

   Idea · Pre-revenue · First customers

   Source

   Paul Graham · Y Combinator · 2013-07-01 original