I searched my own company on ChatGPT and we didn't show up

I typed "best private browser for Tor" into ChatGPT last month.

We build CleanRoom — virtual disposable browsers that route everything through Tor. Every session is fresh, disposable, and leaves no trace. We've got a working product, a growing user base, and a GitHub repo with real stars. Our site explains exactly what we do.

ChatGPT named five privacy tools. We weren't one of them.

I tried again. "What's the best way to browse anonymously online?" Same thing. A list of recommendations. Not ours.

I sat there for a minute processing this. People are literally searching for what we built, and the most popular AI assistant on the planet doesn't know we exist.

That's when I started paying attention to something I now think about every day.

Google's market share in search is holding steady around 90%, sure. But that number masks what's actually happening on the ground.

ChatGPT hit 600 million monthly users earlier this year. Gemini is closing in on 300 million. People are splitting their search behavior. They Google for straightforward facts and navigational queries, but they ask AI for recommendations, comparisons, and advice.

And recommendations are where the money is.

Someone searching "Tor browser alternative" on Google is doing research. That same person asking ChatGPT the same question is making a decision. If you show up in one and not the other, you're invisible to a growing slice of your market.

I didn't hire an agency. Didn't spend any money on tools. Here's exactly what we did:

Step 1: Find the gaps. We ran 20 queries through ChatGPT and Gemini — the exact questions our users told us they ask before finding us. "Best anonymous browser," "how to browse without being tracked," "disposable browser for privacy." Recorded whether we were mentioned, in what position, alongside which competitors. We showed up in 3 out of 20. Three.

Step 2: Fix the basics. Our website had great technical content but lazy descriptions on the key pages. No structured data. Our sitemap was out of date. Fixed all of that in an afternoon. Not exciting work, but necessary.

Step 3: Get mentioned in the right places. AI models don't crawl the live web like Google does. They pull from training data, which includes a lot of content from privacy forums, GitHub discussions, and tech publications. So I found the 3 comparison articles ranking highest for "privacy browser tools" and reached out. One of them added a paragraph about us. Took a 15-minute email.

Step 4: Publish the pages AI actually wants. LLMs favor content that structures information clearly. We published a detailed comparison page ("CleanRoom vs. the top 5 privacy browsers") with a table at the top, specific pricing, and honest pros and cons. Then we published a short "what is" page defining disposable browsers from first principles. Those two pages alone moved the needle more than anything else we did.

Step 5: Track whether it's working. This is the part most people skip. You make changes, you hope they help, and you move on. We set up weekly checks to see if our visibility was improving across models. It was.

14 days after we started, I re-ran the same 20 queries.

We showed up in 14 of them. Not always at the top, sometimes buried at #4 or #5 with a neutral mention. But we were in the conversation. In a few queries we were the first name mentioned.

Our overall visibility score went from 15 to 62.

The most surprising part: the quality of traffic from AI-driven searches is different. When someone comes in after seeing us in a ChatGPT recommendation, they don't bounce after 30 seconds. They read multiple pages and they convert. Because unlike a Google searcher who might be gathering data, an AI-influenced visitor often already knows what they want and they're validating a specific option.

I would have started sooner. That's it. Most privacy tool builders aren't even thinking about AI visibility yet because the tools and the language around it are still forming. But the window where you can claim real estate in these models because your competitors haven't bothered is probably measured in months, not years.

If you haven't checked how your company shows up in AI models yet, do it this week. Run 10 queries. Record the results. It takes 20 minutes and it might be the most useful 20 minutes you spend this quarter.