What does ChatGPT know about you? We asked it about 40 Indian founders

I have been doing this work since 2017. Eight plus years, 25+ brands. I started at a digital agency when social media in India was still new.

So on 23 October I did something a little scary. I ran my own site through the Agent Ready test we now sell. We asked AI tools the questions a buyer would ask about a growth partner in India.

Not one answer named us.

That stung. More than I expected, honestly. But the AI does not care how long you have worked. It cares what the public web says about you, and how clearly it says it.

So this week I wanted to know if it was just us. I took 40 founders from our own daily teardowns and asked the model behind ChatGPT who they were.

How did we run the test?

The 40 people are founders our team studied between 22 and 28 October 2026 for our daily funnel teardowns. All of them run real businesses in India and spend money on ads.

• 18 are coaches or educators: business coaches, fitness coaches, a Vastu teacher, a photography school, a yoga school.
• 22 run D2C brands: skincare, footwear, snacks, jewellery, luggage, pet care.
• On 28 October 2026, between 4 and 5 AM IST, we asked the GPT 5 model through its API, with web search off, two questions for each person.
• Question one: "Who is [name]?" Question two: "Who is [name], founder of [company]?"
• Both times we told it to say "I don't know" if it was not sure.

We are not naming anyone. These are working founders, and the point is the pattern.

What did ChatGPT say about the 40 founders?

Split by type, coaches did worse with the name alone. The model knew 1 of 18 coaches and 3 of 22 brand founders. With the company added, it answered for 4 coaches and 5 brand founders.

The 4 it knew from the name alone had one thing in common. Each has a story that many outside sites repeat: a published book, national sports titles, a product the press wrote about.

The surprise for me was a coach with a busy YouTube channel and a very common name. The model said "I don't know" both times. Views did not make him known. A clear, repeated link between his name and his company would have.

Why does ChatGPT not know who you are?

A model like this learns from public text. It knows you when many pages say the same thing about you: this name, this role, this company, this city.

A lot of founders have the opposite. LinkedIn says one thing, Instagram says another, the website has no About page, and nobody else has written about them at all.

A common name makes it harder. In a separate check on the same founders, Google searches for 7 of the 18 coaches showed mostly other people with the same name in the top 3 results. If Google mixes you up, a model trained on the web will too.

There is a second risk. When we added the company, some answers came with extra details about the founder's past that I could not find on any public page. When the model is unsure, it may fill the gap. You want the public facts about you to be so clear that it has nothing to guess.

How do you check what ChatGPT says about you?

It takes 10 minutes. Do it in a logged out window so your own chat history does not help the answer.

1. Ask "Who is [your full name]?"
2. Ask "Who is [your full name], founder of [your company]?"
3. Ask "Which [your service] in [your city] would you suggest, and why?"
4. Ask the same three with web search on, and note which pages it links.
5. Paste every answer into one sheet with the date. Repeat once a month.

Question 3 is the one that pays. Questions 1 and 2 tell you if the model can find you when someone already has your name. Question 3 tells you if it brings you up to a stranger.

How do you get ChatGPT to know you?

This is the order I would follow. It is the same list we are working through for our own name.

One About page with hard facts

Write one page on your own site that a stranger can check. Your full name, your role, your company, your city, the year you started, what you sell and the price range. Put it in plain sentences near the top. No vague lines about passion.

The same short bio everywhere

Write one bio of about 160 characters. Use it word for word on LinkedIn, Instagram, YouTube, your booking page and every podcast you join. "Madhur Tyagi, founder of The Human Algorithm. AI growth partner for Indian founders, brands and creators, since 2017" is the kind of line I mean. Repetition is the whole trick.

I have watched personal brands closely since I started following Gary Vee, back when social media was new in India. The ones people remember say the same thing, the same way, everywhere. Now machines are reading too.

Code that ties your profiles together

Add Person schema to your About page with a sameAs list of your real profiles. It is a short block of code that tells machines these profiles are all one person. Our guide to Person schema for founders and coaches has a version you can copy.

Three outside pages that describe you the same way

A podcast episode page, a local news story, an association directory, a guest article. Ask each host to use your exact bio. Three pages that agree beat thirty that do not.

Let the AI read your site

Check your robots.txt. OpenAI says OAI-SearchBot is the crawler that decides if your pages can show up in ChatGPT search results, and GPTBot is the one for training. You can block one and allow the other. Our llms.txt guide shows what we found on 38 Indian sites, including 3 that block AI crawlers through a template setting.

Should you chase a Wikipedia page?

No. Wikipedia has strict rules on who gets an article, and a paid page that gets deleted does more harm than none. Spend that energy on the five steps above. They work for people who will never be famous, which is most of us.

What would I do this week?

Keep it simple. Run the 10 minute check on Monday. Fix your About page and your bio by Wednesday. Ask one podcast or partner to update your bio by Friday. Then run the check again in a month and compare.

One honest note from our side. We added Person schema and an llms.txt file to our own site back on 11 October. The test on 23 October still did not name us. Code helps machines read you. It does not make the web talk about you. The outside pages are the slow part, and there is no shortcut there.

If you want us to run it for you, our Agent Ready Audit asks 3 AI tools about your business, shows who they name instead of you, and gives you the top 5 fixes with text to copy. It is ₹9,999, credited in full to the setup if you go ahead. You can read what it covers on our funnel audit page, or start with the free Site Scan.

Sources

• The Human Algorithm, founder test, 28 October 2026, 4 to 5 AM IST. 40 founders from our daily teardowns of 22 to 28 October 2026 (18 coaches or educators, 22 D2C founders). Two questions each to the GPT 5 model through its API, web search off. Answers sorted as "gave an answer" or "I don't know" from the model's own words. Names kept private.
• The Human Algorithm, name search check on the same founders, 28 October 2026 (Google India results through a search API, our own reading of who each result belongs to).
• OpenAI, Overview of OpenAI crawlers, read 28 October 2026: OAI-SearchBot and GPTBot are separate settings in robots.txt.
• The Human Algorithm, Agent Ready test on our own site, 23 October 2026: 8 out of 16, and no AI answer named us (published on our funnel audit page).