Peec AI: Eight LOIs from a Vibe-Coded Prototype
Type: case-study
Stage: Stage 4: Prototype Proof
Difficulty: intermediate
Marius Meiners built a prototype in 1.5 days using Lovable — no co-founder, no production code, no backend. Eight customers signed letters of intent based on it. Antler wrote a €100k check. By 2026, Peec AI had 2,000+ customers and $8.6M ARR. The prototype was disposable by design.
Overview
Marius Meiners built a prototype before he had a co-founder, before he had a production codebase, and before he had written a single line of real backend code. Eight customers signed letters of intent based on it anyway.
The problem thesis
When OpenAI launched search in October 2024, Meiners saw the implication immediately: every brand that had optimized its visibility for Google would need to do it again — for AI. ChatGPT, Perplexity, and Gemini were becoming how people found products and companies. No tool yet existed to track how brands appeared in those results. That was the gap.
The build process
Meiners was four years removed from writing production code. He didn't wait for a technical co-founder. He opened Lovable — a browser-based AI tool — and described what he wanted in natural language. The prototype took one and a half days to build.
It was not production-ready. There was no real data pipeline, no billing infrastructure, no security hardening. But it looked credible, it was interactive, and it demonstrated the core value proposition clearly enough for a customer to understand what they were being shown.
That was the only standard it needed to meet.
The signal
Meiners brought the prototype to eight potential customers. All eight signed letters of intent. Antler wrote a €100,000 check. His CTO joined and rebuilt the product as a production application in six weeks. Peec AI launched in February 2025.
The prototype code was never used in production. It was disposable by design — a functional simulation built to generate commitment, not to ship.
The outcome
By 2026, Peec AI had over 2,000 customers and $8.6 million in ARR. The day-and-a-half prototype cost nothing but time. The letters of intent it generated funded the real build.
The lesson
The prototype's job is not to be the product. Its job is to make the business real enough that strangers will sign their names to it.
Polish is a distraction at this stage. Commitment is the only signal that matters. Eight customers saying yes to a rough prototype is worth more than a thousand saying "that looks interesting" to a polished one.
Build the minimum artifact that can generate a real decision. Then replace it with something that lasts.