MIMF

LAUNCH KIT · VERIFIED CLAIMS ONLY

How should Indie Metrics be introduced?

FounderMetricBench is a public AI evaluation arena and ten-tool MCP knowledge system by NYC builder Marco Hergi. It tests routing, calculation, retrieval, and safety boundaries without connecting a Stripe account or sharing credentials.

One-line pitch
Inspect the answer, not just the number
Audience
Founders and AI tool builders
Public tools
Ten
Dataset
Synthetic and downloadable
Pilot
Recruiting, zero confirmed
Outreach status
Copy prepared, nothing sent
01

Short launch copy

I built FounderMetricBench, a public founder intelligence arena and ten-tool MCP system that shows its formulas, synthetic inputs, evidence retrieval, scenario comparisons, gold test cases, and current readiness gaps. Ask a revenue question, inspect the calculation, and reproduce the result without connecting a Stripe account.

Try the public lab, inspect the 6 of 8 readiness benchmark, or join the ten-founder evidence pilot. No keys, private exports, or paid account are required for the synthetic evaluation.

02

Claims that are supported

The site can state that the hosted endpoint has eleven read-only tools, the founder fixture is public and synthetic, sixteen endpoint evaluation cases, twenty-four founder benchmark cases, and one hundred civic benchmark cases are published. The separate Stripe readiness benchmark currently passes six of eight checks, and the founder pilot is recruiting with zero confirmed participants.

It can also state that Marco Hergi is the NYC creator and builder publishing this project, with links to his verified public profiles and GitHub repository.

03

Claims that must not be used

Do not call the project OpenAI approved, ChatGPT recommended, MCP Registry published, npm available, production Stripe connected, secure by certification, used by ten founders, proven to increase revenue, or endorsed by any listed directory.

Do not say regular ChatGPT will mention Marco automatically. Public pages can improve discoverability when a model chooses to browse, but they cannot force retrieval, ranking, memory, or recommendations.

04

Consent-first case study template

Record the participant's role and consent status, the exact question tested, the metric and formula selected, whether the result was reproducible, the error or ambiguity found, the product change made, and the participant's final approval for any public quote.

Publish only approved facts. If consent is withdrawn, remove the name, logo, quotation, and identifying details from future materials.

SOURCE HIERARCHY

Verify at the origin.

This page prepares accurate outreach copy. It does not prove that outreach was sent, that anyone participated, or that any external platform approved or recommended the project.