When buyers ask AI, do you come up?
Tell us your brand and what you sell. We generate three representative buyer questions (no brand bias), run each one through every engine you have a key for, and a judging model classifies how each answer treats you; recommended, neutral, negative, or absent. One number: your AI recommendation rate, the share of answers that actually recommend you, with named and cited flags alongside.
What is Would AI Recommend You?
Would AI Recommend You? is a free, bring-your-own-key tool that generates three representative buyer questions for a purchase like yours, runs each one through Claude, ChatGPT, Gemini, and Perplexity, and has a judging model classify how each answer treats your brand; recommended, mentioned neutrally, mentioned negatively, or absent. The headline AI recommendation rate is the share of answers that actually recommend you, with named and cited flags shown alongside.
How it works
You enter your brand, domain, and a one-line description of what you sell, plus an API key for any engine. One model first writes exactly three unbiased, representative buying-intent questions (your brand is never shown to it), then every configured engine answers each question with web search or grounding where supported.
A judging model then reads each answer and classifies its stance toward your brand, and the recommendation rate is the share of answers classified as recommending you. Results vary run to run, so treat one run as a snapshot and repeated runs as the trend.
Who it’s for
For founders and marketers who suspect AI assistants are recommending competitors; the outcome is a concrete recommendation rate plus the exact question-engine pairs where you are absent or only neutrally mentioned, which become the priority list for GEO and content work.
In practice
An ecommerce tracking provider enters its brand and one sentence about server-side tracking; the tool generates three representative buyer questions, and the judge finds it recommended in only two of eleven engine answers (neutral or absent in the rest, cited in none); and those misses become the pages and entity work its team ships next.
Your buyers’ questions, every engine’s verdict.
Add a key for any engine (used once, never stored). More engines = a sharper read.
Use your own API keys : used once for your request, never stored on our servers
Keys travel over HTTPS to our server, which calls each AI provider for this one request, then discards them, never logged or stored. Answers vary run to run; this is a snapshot, and tracked runs give the trend. Want it fixed, not just measured? That’s GEO.
Questions, answered.
Yes, the tool itself is free to use. Because it is BYOK, any usage costs from the AI provider tied to your API key are your responsibility.
Yes, you need to bring your own API key. Your API key is sent over HTTPS to NYFTY's server only to make the request(s) for that single run, is never logged or stored, and is discarded immediately after.
It generates exactly three representative buyer questions from your product description, model-generated prompts of the kind real prospects might type, not pulled from your search logs or customer data, and asks each one across the supported AI engines. Your brand is never shown to the model that writes the questions, so the phrasing stays unbiased.
It covers the AI engines currently supported by the tool for that run and breaks results out by engine. Exact model availability can change as providers update access, so the run output is the source of truth for what was checked.
It reports an AI recommendation rate: a judging model reads each engine answer and classifies how it treats your brand, recommended, mentioned neutrally, mentioned negatively, or absent, and the rate is the share of answers that actually recommend you (not just any mention). It also breaks results out by question and engine, with named/cited flags. Use it to spot where AI recommends you, where competitors win instead, and which content or authority gaps to investigate next.
Named means the AI answer mentioned your brand in its response. Cited means the answer used or referenced a source connected to you, which is a stronger signal that your content or authority was part of the answer.