Entry #11 · Apr 14, 2026

Your customers are asking AI about you right now. You just can’t see it. We can.

We’ve spent three months building an AI site for Genymotion, an Android emulator company. The AI site makes their product show up in ChatGPT, Claude, and Perplexity answers. Citation rate went from 14% to 83%. We’ve written ten articles about the technical side: crawler behavior, sitemap structures, bot architectures.

This article is different. This one is about what we’re learning about the people who ask the queries.

Something new is happening

People are evaluating products through AI now. They ask: “What’s the best Android emulator?” “How much does it cost?” “Does it work on my Mac?” They also happen in ChatGPT, Claude, Perplexity.

In that case, conversations often end with a recommendation. The recommendation pushes the user in one direction or another. The company whose product is being discussed has no idea it happened.

When people use AI answer engines, you don’t see page views. Users don’t fill in forms. Google Analytics sees nothing.

The conversation happens. A decision gets made in a place the company can’t see.

We think we are able to see a significant part of that traffic.

Two windows into the same audience

We have two data sources for Genymotion. Together they give a view full of insights that we’re sharing here.

The Rozz chatbot

The Rozz chatbot sits on Genymotion’s website. The Rozz chatbot captures what visitors ask.

The Rozz chatbot is an answer engine that knows the website, help center, and product documentation. 667 conversations in March represent over 1,300 questions.

Each conversation records what the person wanted to do. Each conversation records what problem the person had. Each conversation records whether the person got an answer.

This is direct. The user types a question. The chatbot answers.

The AI site

The AI site sits at rozz.genymotion.com. The AI site serves structured content to AI platforms.

When someone asks ChatGPT about Genymotion and ChatGPT looks up a page from the AI site to answer, we see it in the CloudFront logs.

We don’t see the user’s question directly. We see which pages were fetched. The pages are Q&A pages generated from real chatbot conversations.

The page titles are the questions themselves. By clustering fetches by timing and IP, we can reconstruct multi-turn sessions.

We’re not saying that every ChatGPT conversation about Genymotion hits the AI site. Many are very likely answered from training data alone or via the index.

We still see a lot of queries. 3,830 in March are grouped into over 2,500 sessions.

We see the conversations where ChatGPT needed to look something up. Those tend to be the specific, current questions: pricing, compatibility, recent releases.

What we’re seeing in the ChatGPT sessions

This week, ChatGPT fetched 1,323 pages from the AI site during approximately 500 live user sessions.

Here are some of the sessions we reconstructed.

A pricing evaluation from Madrid.

Eleven pages are fetched across five rounds. About five minutes total pass.

The session starts with root access topics. The session then moves to the free personal use Q&A. The session then includes the SaaS vs Desktop pricing comparison.

The session then moves to the full pricing breakdown. The page sequence tells a story. Someone explores the product. Someone then asks “can I get it free?” Someone then asks “ok, what does it actually cost?”

A 22-minute session from the US West Coast.

Four rounds happen. A 17-minute gap happens between rounds two and three.

The last page fetched is the pricing Q&A. The session ended on the purchase question. The pages do not reveal what happened during those 17 minutes.

The same macOS question, twice, from two continents.

Two independent sessions happen on the same day. One session is from the US. One session is from Europe.

Both sessions fetched the “Is Genymotion available for macOS?” Q&A page. This question appears in the logs daily.

ChatGPT doesn’t seem to know the answer from training and looks it up every time.

Three VirtualBox bug reports on the same day.

Three separate sessions happen. Three different regions are involved.

All sessions ask about the same problem. The problem is “I upgraded VirtualBox and Genymotion no longer works.” The sessions could be an interesting product signal for the support team.

A CLI runbook session.

Eleven pages are fetched across four rounds.

The user pulls both the gmtool and gmsaas CLI runbooks. The runbooks are the step-by-step command references added to the AI site specifically for developer tooling workflows.

More usage from coding tools is expected. The goal is to develop the runbooks as a sales channel.

What 667 chatbot conversations reveal

The ChatGPT session data shows which questions the AI is looking up.

The chatbot data on the website shows who is visiting. The chatbot data also shows what visitors want to do.

Genymotion’s marketing targets enterprise buyers. Genymotion’s marketing targets CI/CD automation. Genymotion’s marketing targets mobile security testing. Genymotion’s marketing targets cloud deployment at scale.

Here is what people actually told the chatbot they wanted to do in March: | User goal | Conversations | | --- | --- | | Use Genymotion Cloud | 14 | | Get a free license key | 9 | | Know prices and billing plans | 9 | | Install on Windows step by step | 7 | | Set the language to Chinese | 6 | | Install on PC to use Instagram | 5 | | Play Minecraft on Steam | 3 | | Run TikTok on PC | 3 | | Download eFootball | 2 | | Update WhatsApp | 2 |

And the enterprise-buyer goals?

| User goal | Conversations | | --- | --- | | Create an API token in the SaaS portal | 4 | | Perform native MITM with TLS interception | 4 | | Run automated UI tests | 2 | | QR provisioning for company-owned device | 1 |

Out of 667 conversations, fewer than 20 show clear commercial intent. The chatbot has a 95.5% satisfaction rate and 88.7% resolution rate.

The chatbot works well. The chatbot mostly answers questions from hobbyists and individual users trying to run mobile apps on their PC.

The enterprise audience the marketing team targets represents less than 5% of the conversations.

The website analytics would show page views. The chatbot shows intent.

What we think this means

There are two things happening here. The two things matter beyond this case study.

AI is THE discovery and evaluation channel, and it’s invisible to most companies.

The pricing evaluation from Madrid happens. The late-night deliberation from the US happens. The macOS compatibility checks happen daily from multiple continents.

These are real product evaluations. Some evaluations lead to purchases. Some evaluations do not lead to purchases.

The company has no visibility into these evaluations without an AI site generating logs.

This is new. A year ago, these conversations would have been Google searches that showed up in analytics.

Now these conversations are AI conversations. These conversations might not lead to any traffic on a human site. The AI site can still notice them, at least in part.

The combination of chatbot + AI site creates a feedback loop that a website alone can’t.

The chatbot tells who is visiting the website. The chatbot tells what visitors want to do.

The AI site tells what questions AI platforms are looking up about a product. Together they reveal what audience is actually showing up.

The actual audience might not be the one the marketing team targets.

What we’re trying next

Now that conversations can be seen, the question becomes: can the conversations be influenced?

If 95% of AI-mediated conversations are about free usage and mobile gaming, and 5% are about enterprise CI/CD, a ratio shift is questioned.

The ratio shift question is whether the AI site can prioritize business use cases. The prioritization would make AI platforms more likely to recommend Genymotion for enterprise workflows.

The prioritization would also reduce recommendations that only confirm Genymotion exists for hobbyists.

We think so. We’re working on it now. That’s the next article.

Get this for your company

Rozz gives visibility into the AI conversations happening about a product. Rozz also provides tools to influence what AI recommends.

AI site + chatbot + analytics are combined. One infrastructure turns invisible AI conversations into a measurable channel.

$997/month | AI site + chatbot + analytics → Book a call | → See how it works | → rozz@rozz.site

Entry #10: Three Citation Bots. Ten Articles.

All Entries

→ Entry #12: AI Site Structure Matters More than We Thought

> Data source: Rozz chatbot conversation logs on Genymotion.com and CloudFront access logs for rozz.genymotion.com, March 1 – April 14, 2026. ChatGPT sessions reconstructed by clustering page fetches on IP and timing. Intent classification from the chatbot’s own conversation analytics.

> Author: Adrien Schmidt, CEO, ROZZ

Serial tech entrepreneur with 10+ years experience building AI systems including Aristotle (conversational AI analytics) and products for eBay and Cartier. Previously founded Squid Solutions and built AI products like Aristotle, the conversational big data analytics chatbot, and an AR jewelry try-on device for Cartier.

April 14, 2026 | Data period: Mar 1 – Apr 14, 2026

Entry #11 · Apr 14, 2026