Frequently asked questions

Boss.cc is a free tool that evaluates your startup or product idea using five different AI models. Each model scores a different angle — viability, market, competition, monetization, and risk — and a synthesis judge returns a single GO, PIVOT, or NO-GO verdict, usually in about ten seconds.

Yes, it is free and there is no signup or credit card required. You paste your idea, click to confirm you are human, and run the panel. The only limit is 20 submissions per hour per IP address to keep bots out.

A single AI model tends to agree with you and has one set of blind spots. Five models from four different labs (Meta, Qwen, NVIDIA, Mistral) each judge a different dimension and never see each other’s answers, so they can’t fall into groupthink. When they agree, that is strong signal; when they disagree, the split shows you where the real risk is.

GO means the idea scored strongly across the panel and is worth building now. PIVOT means there is a real seed but the current shape is wrong — the verdict points at a better angle. NO-GO means the panel found likely killers within 6–12 months, so you should save your runway.

Each of the five panel agents scores your idea from 0 to 10 on its dimension, with written reasoning. A separate synthesis model weighs all five verdicts to produce the final score and recommendation. You can read every individual verdict, not just the summary.

You choose. By default a verdict is public and appears in the idea catalog with a shareable link. If you mark it private, it is hidden from the feed and reachable only through the unique URL we generate for you.

Anything you can describe in a few sentences — a business venture, a side project, a content or article idea, a product, and more. You pick the evaluation lens when you submit, and the panel adapts its judging to that purpose.

The panel runs on Meta Llama-4 Maverick, Qwen 3-next 80B, NVIDIA Nemotron (reasoning), Mistral Medium 3.5, and NVIDIA Nemotron Super 49B, with a Llama 3.3 70B synthesis judge. Each agent has a fallback chain, so a single model outage does not break your verdict.

Usually under fifteen seconds. Each agent posts its verdict the moment its model returns, so you watch the panel fill in live rather than staring at a spinner.

Treat it as a fast, brutally honest first opinion — five of them, actually — not a substitute for talking to real customers. The value is in catching obvious problems early and in the disagreement between models, which surfaces risks a single advisor would miss.