How to make ChatGPT know who you are
By the Rashahni team, Cairo · Updated 16 November 2026
How do I make ChatGPT describe me correctly?
Rashahni’s short answer: ChatGPT describes you from what is published about you, so you fix the sources, not the chat. Use one spelling of your name, tie it to your profession and city everywhere, publish one page with your facts, make LinkedIn and your other profiles say the same thing, and re-check the answer from a logged-out browser.
Why does ChatGPT get people wrong?
AI assistants are most accurate about famous people and least accurate about everyone else. Published research found the same pattern: language models answer questions about well-known people far better than about little-known ones (Mallen et al., 2023; Min et al., 2023). To an AI model, most consultants, coaches and founders are little-known people.
In our own check in November 2026, 17 of 30 personal names came back wrong or empty in ChatGPT, while company names did much better. It was one question per name from a logged-in account, so read it as a direction, not a statistic.
The most common failure is not a false fact. It is a namesake: someone better known with the same or a similar name takes over the answer. In the cases we checked, when an engine found the person’s own site, the description was mostly right. The hard part is getting the engine to pick the right person first.
Where does ChatGPT get its answer about you?
From two places, and they change at different speeds.
- Model memory: what the model learned in training. It changes only when a new model is released, which takes months. Nothing you publish today edits it.
- Live search: the pages the assistant opens while it answers. ChatGPT, Gemini, Perplexity and Copilot all do this for many questions about people. This part can change within weeks once the pages it reads are fixed.
Most of what an assistant reads about a person sits on other sites: LinkedIn, directories, professional associations, interviews and news. Large public studies of AI citations list LinkedIn among the most-cited sites. Your own site matters as the place that states your facts plainly, but it rarely decides the answer on its own.
How do I make ChatGPT know me, step by step?
Eight steps. None of them needs special software; they need time and consistency.
Ask first, from a clean browser.
Open a private window, stay logged out or use a temporary chat, and ask three questions: your name alone, your name with your profession, and your name with your city. Ask in Arabic and in English. A logged-in account with memory can steer the answer toward you and hide the problem. Write down what is wrong and which pages it cites.
Choose one spelling of your name.
One in Arabic and one in English, and use exactly those everywhere. If you are Mohamed on one site and Mohammed on another, an engine can split you into two people or merge you with someone else.
Write a facts sheet.
Your name in both scripts, your title, what you do in one sentence, your city, how clients reach you, and a short, a medium and a long bio. Every profile you touch copies from this sheet.
Publish one page you control.
A personal site, or a page on your firm’s site, that states those facts in plain sentences in Arabic and English, with structured data (Person schema) that links your official profiles. Check that its robots.txt lets search engines and AI crawlers in.
Make LinkedIn say the same thing.
A headline with your profession and city, the same bio and the same spelling. Then do the same on the directories and professional associations of your field, and on a Google Business Profile if you receive clients at an office.
Ask for corrections at the source.
If a page says something wrong about you, ask its owner to fix it. Telling ChatGPT it is wrong inside a chat does not change what it tells other people.
Earn real mentions.
An interview, a talk, a guest article, a listing in your professional association. Pages written by others that repeat the same facts are what convince an engine it has the right person.
Ask the same questions every week.
Answers change from one run to the next, so ask several times and count how many answers are right. One screenshot is not a measurement.
How long does it take for ChatGPT to update what it says about me?
Answers that come from live search can change within weeks after the sources are fixed. What the model remembers changes only with a new model, which takes months. Anyone who gives you one fixed date for both is guessing.
What does not work?
- Telling ChatGPT about yourself in a chat. Memory is saved for your own account only; nobody else sees it.
- Buying an “AI ranking” or a guaranteed recommendation. Nobody controls that, and the same question rarely returns the same list twice.
- Fake reviews, invented awards, or stuffing your name with keywords. Engines and people both read them as noise, and a false claim repeated by AI is worse than no answer.
- An llms.txt file on its own. It does no harm, but Google itself says its AI features need no special file or markup.
Can anyone guarantee that ChatGPT will recommend me?
No. Recommendation lists change from run to run: a public study that repeated the same prompts dozens of times got the identical list less than 1% of the time. What can be measured is accuracy, meaning whether the assistant describes you correctly when someone asks about you by name. That is the part you can work on.
Do I need a service to do this?
No. You can do every step above yourself; it takes time more than skill. Rashahni (رشّحني) is a service run by a team in Cairo that does this work for service providers in Egypt, the Gulf and the United States. If you first want to see what ChatGPT and Gemini say about you today, request the free AI Name Report Card from your country page.
Questions
Does ChatGPT know who I am?
Ask it from a logged-out browser: your name alone, then your name with your profession and city. If it names someone else, gives an old job or says it does not know, the sources it reads about you need work.
Can I edit what ChatGPT says about me?
Not directly. There is no form that rewrites an answer for everyone. You change the pages it reads: your site, LinkedIn, directories and the pages that mention you.
Does my own website help?
Yes, as the place that states your facts in plain words. Most answers about people also draw on other sites, so your profiles and mentions need to say the same thing.
Is this the same as SEO?
It stands on the same foundations. SEO aims at your position in Google results; this aims at the sentence ChatGPT and Gemini say about you, and it is measured by asking them.
Does this work for Arabic names?
Yes, and spelling matters even more. Choose one Arabic and one English spelling, use them everywhere, and check the answer in both languages.
Sources
- Mallen et al., “When Not to Trust Language Models”, ACL 2023 · arxiv.org/abs/2212.10511
- Min et al., “FActScore”, EMNLP 2023 · arxiv.org/abs/2305.14251
- Google Search Central, “AI features and your website” · developers.google.com/search/docs/appearance/ai-features
- OpenAI, “Overview of OpenAI crawlers” · platform.openai.com/docs/bots
- SparkToro and Gumshoe, study of how often AI tools repeat the same recommendation list (2,961 prompts, run 60–100 times each)
- Rashahni, own check of 30 personal names in ChatGPT, November 2026 (one question per name, logged-in account)
Rashahni is independent and is not affiliated with OpenAI, Google, Microsoft, Anthropic or Perplexity.