When a buyer asks ChatGPT for a recommendation,
is it you or your competitor?
Search is moving into AI answers. We ask ChatGPT, Gemini, Perplexity and Claude the exact questions your buyers ask about your category, record every answer verbatim, and show you question by question where you are absent, who gets recommended instead, and the three fixes to start with. No estimates, no proprietary "AI score": the answers themselves, dated.
Request an audit Book a 20-minute callA recorded snapshot
40+ buyer-style questions per audit, asked in each assistant's own app, answers stored word for word with the date. Share of voice, missing questions, competitor mentions.
Where you leak buyers
The questions where assistants name a competitor and not you. Which competitor, how often, and in what position.
Fixes, page by page
Source pages that answer the missing questions, comparison pages, entity consistency across your site and listings. Concrete, prioritised, written for your team or agency.
How it works
Tell me your category and competitors
Your brand, the category your buyers use in plain words, your market, and up to five competitors you care about. That is the whole brief.
I ask the assistants and record
Discovery, comparison, local, pricing and trust questions, asked in ChatGPT, Gemini, Perplexity and Claude. Every answer saved verbatim with a timestamp; the report links to them.
You get the audit in 5 working days
A web report plus a PDF: share of voice by assistant, the question-by-question table, the competitors that replace you, and the three to ten fixes in priority order. Optional monthly re-run so you can see the needle move.
Pricing
AI Visibility Audit
- 40+ buyer questions, 4 assistants, answers recorded verbatim
- Share of voice, missing-question list, competitor map
- Prioritised fixes with page-level briefs
- Delivered in 5 working days, 45-minute walkthrough call
Audit + Monitoring
- Everything in the audit, then a monthly re-run of the same questions
- Month-on-month share-of-voice trend per assistant
- New competitor alerts and new missing questions
- Monthly fix list, cancel any month
Agency / multi-brand
- Up to 5 brands or client accounts
- White-label reports with your agency's branding
- Shared question bank per vertical
- Priority turnaround
50% to start, 50% on delivery for the audit; monitoring billed monthly. Card or bank transfer (USD, EUR, GBP). Invoice provided.
Sample snapshots
Each sample is a real, recorded snapshot for a real company, prepared from public information so you can judge the actual output. They are labelled as samples, built without the company's involvement, and removed on request.
First samples are being cut now.
Who does this
Gabriel Homoky, founder. I run Automating.hu, an AI automation studio in Hungary that builds AI systems for businesses. AI Visibility Audit is the studio's measurement line: a fixed, repeatable method where a person runs every question, reads every answer and writes every recommendation. Nothing in the report is generated without being checked.
Based in Europe, working with US, UK and Australian companies over email. Reply times are within a business day.
Questions people ask
Is this an "AI score" or a ranking tool?
No. Monitoring tools give you a number from a sampled API. This is a human-run audit of the real consumer apps, with the answers recorded so you can read them yourself. The trade-off is that it is a dated snapshot, not a live dashboard - which is also why the monthly re-run exists.
Can you guarantee we will show up in ChatGPT?
No, and nobody honestly can. Assistant answers change by session, location and model version. What I can show is where you are absent today, why the competitor is cited, and the concrete pages and consistency fixes that give assistants something to cite. The re-run shows whether it moved.
How is this different from SEO?
It overlaps, but assistants weight things differently: direct answers in the first lines of a page, consistent entity data across the web, comparison content and third-party mentions. Many well-ranked sites are absent from AI answers. The audit tells you if that is you.
Is the work AI-assisted?
The questions are asked of AI assistants by definition, and I use AI tools to tabulate mentions. Every answer is read and every recommendation written and checked by me. Reports say "AI-assisted, human-reviewed".
What do you need from me?
Your category in your buyers' words, your market or city, and your competitors. Access to nothing. Five working days later you have the audit.