I run 25 real prompts in ChatGPT, Perplexity and Google AI Mode by hand and send you a white-label report with a plan within 48 hours. Free while I build the product.
One of 25 prompts from a real audit. Client name changed. This is what a buyer in Germany sees — and the client isn't there. The audit shows why, and what to do about it.
A 6-page PDF: where your client appears across 25 prompts, who is cited instead, and five tasks ranked by expected effect. Your logo on the cover, mine nowhere.
I'm building software for this. But before I write code, I'm doing the work by hand: running every prompt, reading every answer, writing the plan myself. It's the only way to build a product that solves the real problem, not the one I imagine. That's why it's free, why I take five agencies at a time, and why every request matters to me.
Two minutes in the form below. I reply the same day to confirm the prompts I'll run.
In the real interfaces, from an EU, US or UK location. I save every answer and every cited URL.
I take five audits at a time. If the queue is full, I tell you the real delivery date in my first reply.
The manual audit is the first version of a product for agencies: the same report, every Monday, for every client. This is the screen I'm building now.
Want to be one of the first agencies on it? Write to Sergei
Delivered within 48 hours. No card, no call required.
Senior Product Manager, 15+ years in software, started as a developer. Since spring I've been running hundreds of prompts in ChatGPT, Perplexity and Google AI Mode to understand why some brands get recommended and others don't.
This audit is how I turn that into something an agency can hand to a client.
I do every audit myself. I read all 75 answers, I check every source, and I write the five tasks with your account manager in mind. If something in the report is unclear or wrong, email me — I answer within a day.
— Sergei, sergei@getgeosay.com
Use them. They track numbers through APIs; I run the prompts in the real interface, where the answer a buyer sees often differs, and then read every answer and tell you what to change. If you already export data from Peec or Semrush, I'll use it as input.
From the client's category, market and competitors: the questions a buyer would actually ask before a shortlist ("best X for Y in Germany", "X vs Z", "is X GDPR-compliant"). I send you the list before running it, and you can swap any of them.
ChatGPT, Perplexity and Google AI Mode. Nothing else for now.
No. The PDF carries your logo and nothing of mine. Send it as your own work.
I'm building a product for agencies and need real cases. After the audit I'll ask for 25 minutes of feedback, and I may offer to keep the reports going. You can say no to both.