Context
The founder of a fintech startup. Not a developer himself, but a technical user — a lifetime in banking product management. Over three weekends he assembled a working prototype with an AI assistant (vibe-coding) — he wanted to show an investor that the idea wasn’t just “words.” The prototype even made it through a first live demo. There’s no team, and he hasn’t hired a freelance developer yet.
The pain
The investor is interested and asks him to “show it isn’t a cardboard box” — he wants someone technical to take a look. The founder doesn’t know whom to trust: the developers he knows immediately propose “rewriting everything from scratch” (that is, selling him a 3-month contract). He can’t judge for himself whether things are actually bad in there, or whether this is just the usual “anything I didn’t write is bad.” His biggest fear — that the prototype has a security hole that means it can’t be shown to the investor or to the first customers.
Approach
Seven days. The first two — 90-minute conversations: we walked through exactly what the prototype does, which data it touches, where the keys and passwords live. The next three days — a detailed technical review of the code and infrastructure: security (where the keys are, how user passwords are stored, whether the test data is properly isolated), the realism of a production launch, an assessment of the debt. One day — writing the report in plain human language: 3 critical issues blocking a show to the investor; 5 medium ones to fix before the first customer; a list of what’s good and what’s not worth rewriting.
The founder got a 7-page document + a 30-minute conversation about it. In the document — clear recommendations on what to do himself (3 items, half an hour each), what to hand a junior developer for 2 days, and what genuinely should be reworked before launch.
Result
The investor got a short version of the report and agreed to the next round of conversations. The founder didn’t sign a 3-month “rewrite everything” contract — instead he hired one part-time developer for 6 hours a week to handle the critical items. The budget for this first quarter: ~$3K instead of the ~$15K his contacts were proposing. Most importantly — the founder gained the calm understanding that his prototype isn’t “cardboard,” but does have 3 specific weak points to cover before the first real user.
What it taught
Vibe-coding prototypes get “rewritten from scratch” more often not because they’re bad, but because a person without a tech director can’t tell “critical debt” from “stylistic differences.” An independent 7-day look draws that line clearly — and often saves the founder the $10–15K they’d have paid for unnecessary rewritten code.
This is an illustrative composite pattern from 8 years of work in IT — not a specific client. AG is in a validation phase.