GTMagic
GTMagic helps small-business owners find out who will buy from them and what to say, using research backed simulated customers
Most small businesses pick their customer and their message by gut feel, and they find out they guessed wrong only after the ad money is gone. Think of a bakery adding catering, a local gym repositioning, or a two-person app company. The owners are busy, they aren’t marketers, and a chatbot doesn’t help much: it hands back a confident, generic answer with nothing behind it. We tested this on a real founder’s pivot, our team lead’s fitness app. A one-prompt chatbot answer scored 5–8 out of 18 against where that founder actually landed, and it never found the customers who buy today. The founder’s own website scores 15 out of 18 on the same rubric. GTMagic is three Claude skills that run as one guided flow, plus a chat-style web prototype. First, a roughly 30-minute intake conversation takes in your website, past ads and their results, and customer reviews. It researches your market, logs every source, and gives you a plain-English brief: your customer groups, the exact words real customers use for the problem, who to go after first, an honest check on your own guess, and the most you can afford to pay per customer. Next, it writes headline options from those real customer phrases and screens them, along with yours, against 12 principles learned from 223 real headline A/B tests. Finally, your current headline goes up against the alternatives and a deliberately generic decoy. Simulated buyers built from your customer groups and real quotes see the options blind, and any of them can pick “none of these.” Code counts every choice, the model only explains the result, and you leave with a cheap real-world test to run this week. Every real quote is checked word for word against its source, and every simulated number is labeled “simulated.” To set it up, a business adds the three skills to Claude (claude.ai with web search and code execution turned on, or Claude Code), says “who should I target?”, and answers the questions. There’s nothing else to install, nothing is stored on our side, and the owner keeps a save file to bring back before every campaign. Here’s what’s still rough. The simulated-audience step isn’t live yet, so the demo runs on clearly labeled fake responses to show the scoring and report end to end. The web app is a clickable prototype. GTMagic’s own run on the founder case hasn’t been scored yet. And our learned headline principles roughly matched plain Claude Opus at picking real winners on 400 held-out tests (47.0% vs 45.3%) without beating it, so we ship them as a checklist, not a predictor. The demo business is fictional, and simulated buyers give you hypotheses to test, not proof.
⏺ Read(summary.md)
It interviews you about your business and researches your market, then works out which customer groups to go after and what to say to them. It tests your current headline against alternatives on simulated customers built from real customer quotes, and suggests a cheap real-world test to run this week.
Best for: Small businesses and founders without marketing staff, such as a bakery adding catering or a gym repositioning, who are choosing a target customer or ad message and don't want to guess.
Summary by Claude from the team's submission and README. Find more free AI tools for your business →