When an investor, customer or journalist asks ChatGPT, Claude, Gemini or Perplexity about a founder, the assistant answers either from what its model absorbed in training or from a live web search of the company's About and team pages, the founder's public profiles and press coverage. If those pages are thin, out of date or disagree with each other, the answer comes back vague or wrong, and the fix starts on the pages the founder already controls.
The question usually arrives in plain words: "Who is the founder of this company?", "What did she do before?", "Is he credible on supply chain software?" The assistant does not call you to check. It works with whatever it remembers or can find, and it writes a confident paragraph either way.
Where does an assistant get what it says about a founder?
There are two routes, and the same assistant can take either one, depending on the question.
The first is memory. A model learns from public text collected before its training ended, so what it says about you without searching reflects what was public at that point. For a founder whose company, title or focus changed since then, that memory describes an older version of them, or nothing at all.
The second is a live search. The assistant runs a web search, opens some of the results and writes its answer from them, usually with links underneath. OpenAI's crawler documentation describes the pieces for ChatGPT: OAI-SearchBot surfaces websites in ChatGPT's search features, and sites opted out of it are not shown in ChatGPT search answers, though they can still appear as navigational links; a separate agent, ChatGPT-User, may visit a page when someone asks ChatGPT a question. Google describes a similar pattern for its own AI features: AI Overviews and AI Mode may use a "query fan-out" technique, issuing multiple related searches across subtopics and data sources, and to appear as a supporting link a page must be indexed and eligible to be shown in Google Search with a snippet (page last updated on 10 December 2025).
So a page about you only helps a searching assistant if the search reaches it, and being indexed is not the same as being found. On 19 October 2026 we asked assistants for the exact titles of 18 articles, each older than three weeks, published on three client sites. ChatGPT found 1 of 18; Claude found 7 of 18. Google had indexed 29 of 30 articles of one of those sites. Being in Google's index did not mean an assistant could find the page. The same mechanics explain why an assistant names your competitors but not your company.
Which pages does an assistant rely on for a founder?
When it searches, an assistant works with the pages that come up for your name and your company's name. For most founders that is a short list:
- The company's About page and team page. Usually the most direct statement of who founded the company, when, and what each founder does.
- A personal site or a founder page, if one exists.
- Public profiles. LinkedIn's help says that some sections of your profile may appear off LinkedIn when people search for you using search tools, like Google or Bing, and that you can opt out or decide which sections are eligible to appear through your public profile. What appears there is the public version you configured, so check what it shows.
- Press and third-party pages: interviews, podcast show notes, conference speaker pages, investor portfolio pages and startup directories. They describe you in someone else's words, and they may still carry an old bio you once sent them.
- Your own posts and articles, where they come up for the topics you work on.
Which of these a given answer uses changes from question to question and from one assistant to the next. When the assistant searched, the links under the answer show which pages it drew on. That list is the most useful thing on the screen, because it tells you where the answer came from.
Why does a thin or inconsistent bio lead to vague or wrong answers?
An assistant writes a paragraph whether or not it has good material. If the material is thin, the paragraph is thin: "the founder of a company focused on innovation" tells an investor nothing, and a one-line team-page bio rarely supports more.
Inconsistency does more damage. The usual conflicts look like this:
- The About page says co-founder and CEO, LinkedIn says founder, and a podcast page still shows a previous company.
- The founding year differs between the website, a directory listing and an old press release.
- The company was renamed, and older pages use the old name without connecting it to the new one.
- Another person with the same name works in a nearby field.
Faced with pages that disagree, the model has to pick one version or blend them, and nothing forces it to say it is unsure. A complaint filed with the Austrian data protection authority shows what can happen when an assistant lacks a fact about a person. According to noyb, when ChatGPT was asked about a public figure's birthday, it repeatedly provided incorrect information instead of telling users that it did not have the necessary data, and OpenAI refused his request to rectify or erase it, saying it could filter or block data on certain prompts but not correct it (noyb, 29 April 2024).
For a founder, the practical lesson is that correcting an answer after the fact is hard. Giving every searching assistant the same clear facts on every page it might open is the part you can do.
What can a founder fix on pages they control?
Start with one canonical bio and make every other page agree with it.
- Write the facts down once. Your full name as you want it used, current role, company name (and its former name, if it changed), the year you founded it, what the company does in one plain sentence, two or three earlier roles, where you are based, and the two or three topics you want to be asked about. Write sentences an assistant could quote, not slogans.
- Give yourself a page. A founder page on the company site, separate from a grid of photos on the team page, gives a search one address that is about you. Google's ProfilePage documentation says the primary focus of such a page must be a single person or organization affiliated with the overall website, lists an employee page on a company website as an example, and recommends a sameAs property with the URLs of your other external profiles (page last updated on 8 September 2026). The markup is optional: Google's AI features page says there is no special schema.org structured data you need to add. The text comes first.
- Make the profiles match. Update your LinkedIn headline and About section, your X bio and any personal site to the same role, company and topics. In LinkedIn's settings, check which sections your public profile shows.
- Chase the old bios. Search your own name and send the current bio to the conference, podcast and directory pages that still carry an old title or company. Where a page uses your company's former name, ask for the new one to be added.
- Check that the pages can be reached. If your website blocks the crawlers that search tools rely on, a searching assistant may not find it at all; OpenAI, for one, states that sites opted out of OAI-SearchBot are not shown in ChatGPT search answers. Ask whoever runs the site to confirm that the About and founder pages are not blocked.
- Publish on your topics. If you want to be described as an expert in something, there have to be pages where you say something specific about it. How to get ChatGPT to recommend you as an expert covers that side, and how to post consistently on LinkedIn as a founder covers the habit. If writing is the bottleneck, our separate LinkedIn and X product drafts posts in your own voice on topics you choose, and you approve or edit every one.
How do you check what the assistants say about you?
Ask the questions the people who matter to you would ask, in the words they would use:
- "Who is [your name]?" and "Who is [your name], founder of [company]?"
- "Who founded [company]?"
- "What did [your name] do before [company]?"
- "Is [your name] an expert in [your topic]?"
Put them to ChatGPT, Claude, Gemini and Perplexity, each in a new chat, and ask each one more than once. Answers vary. Asked the same question three times, an assistant often names different businesses; we count a business as named for a question if any of three asks names it, and compare checks only on the same questions and the same number of asks. A single screenshot proves little.
For each answer, note three things: what it got right, what it got wrong or left out, and which pages it linked. A wrong fact with a linked source tells you which page to fix or whose editor to write to. A wrong fact with no link gives you no page to point to; it may come from the model's memory. Fixing your pages does not rewrite that memory, but it gives the assistant something better to find the next time it searches. After you change your pages, ask the same questions the same number of times, so you compare like with like.
How to check if ChatGPT, Claude, Gemini and Perplexity name you walks through the same method for a company. If you run one, ItGrows measures whether ChatGPT, Claude, Gemini and Perplexity name it, with one guarantee: in 90 days, AI assistants will name you more often than on launch day — or we work for free.
Frequently Asked Questions
Does ChatGPT read my LinkedIn profile?
Do not assume either way. LinkedIn's help says some sections of a profile may appear off LinkedIn in search tools such as Google or Bing, and that you decide which sections are eligible through your public profile. When an assistant searches, the links under its answer show whether your profile was among the pages it used, and your own website remains the one place whose content you fully control, so keep the full bio there too.
Do I need schema markup on my founder page?
Not to be eligible for Google's AI features: Google says no special schema.org structured data is needed for them. Google documents ProfilePage markup, with sameAs links to your other profiles, for pages such as an employee page on a company website; it is optional, and clear text with consistent facts comes first.
Can I tell ChatGPT that it got my details wrong?
You can, but a correction in your own chat does not change what it tells an investor in theirs. In the case noyb described on 29 April 2024, OpenAI said it could not correct a wrong date of birth, only filter or block data on certain prompts. Fixing the pages that assistants find when they search is the route you control.
Will a better bio make assistants recommend me as an expert?
No single answer is guaranteed to name or recommend a particular person. A clear, consistent bio makes it more likely that what an assistant says about you is accurate when it does mention you. Being named for a topic also depends on what you have published about it and what others write about you.