01
Access
If the answer engines cannot fetch your pages, nothing else you do exists.
robots.txt, bot protection, noindex and nosnippet, JavaScript-only rendering. Most blocks are accidents nobody ever sees.
Generative engine optimization and AI SEO services.
Get named when buyers ask ChatGPT, Perplexity, Gemini or Google AI who to hire: generative engine optimization and AI SEO, measured before and after.
AI search optimization is the work of getting a business named in the answers that ChatGPT, Perplexity, Gemini and Google’s AI features write. SEO is being found in a list of links. AEO, answer engine optimization, is being the answer to a question. GEO, generative engine optimization, is being named inside an AI-written answer.
When someone asks ChatGPT, Perplexity, Gemini or Google’s AI who to hire, the answer names a few businesses and stops. There is no second page. Either you are one of the names, or the enquiry goes to whoever is.
The engines decide in three steps: can they fetch your pages, do those pages state plainly what you do, and can they corroborate it elsewhere. Every step is inspectable and every step has a fix. That is the work; the strategies behind it are in how to improve brand visibility in AI search engines.
How the work runs
- [01]ScanNeshatScan on the pages that matter: home, the service pages, the location pages. Every blocker and warning with its evidence. This is free and takes minutes.
- [02]Accessrobots.txt corrected for the answer crawlers while keeping any training opt-out you want. noindex and nosnippet removed where they should not be. Bot protection checked at the edge. JavaScript-only pages made readable.
- [03]UnderstandingEach core page rewritten to answer the buyer's question in its first sixty words, with the facts an engine needs: what, for whom, where, how much. One schema graph that says the same thing the page says.
- [04]TrustOne business name everywhere. Profiles and directories that corroborate the site. Sources that mention you for the questions you want to be named for. Founder pages with a face and a record.
- [05]MeasureThe same buyer questions asked of the same engines, signed out, on a schedule. Named or not named, dated. That is the only metric that matters and the only one I report.
What you get each month
- Citation tracking on agreed buyer questions, captured and dated
- Two answer-first pages a month, written to be quoted
- Entity and profile building across the sources AI engines weigh
- Technical monitoring with re-scans after every change
- 30-minute monthly review
Three months minimum, because engines re-crawl, re-index and re-weigh on their own schedule and the citation captures need time to move. After that, month to month.
What I will not promise
Nobody can guarantee that an AI engine will mention you, and anyone who does is selling you a number they made up. What I guarantee is narrower and checkable: every access blocker cleared and re-scanned, every page stating its facts, every claim on the report backed by a dated capture. If the engines still do not name you, you will know exactly what they saw and why.
Who this is for
Owner-led businesses whose buyers ask questions before they buy: law firms, medical and dental practices, contractors, agencies, B2B software, specialised local services. Especially businesses that serve a community in a second language, where the engines have very little to go on and a clear site wins quickly.
It is not for sites that cannot change their pages, for businesses without a real address or a real person behind them, or for anyone who wants a report instead of a result.
Where to start
Run the scan on your own site. If it shows blockers, the one-time fix clears them in five business days. If it is clean and you still are not being named, that is the monthly work. To watch whether the assistants name you, week by week, there is AI visibility tracking. Either way the first step is the same.
How this works
[1]how AI decides who to name
Three layers. Every one of them is checkable.
An AI answer is not magic and it is not a popularity contest. It is a fetch, a reading and a judgement. Each step can be inspected, and each has a fix.
02
Understanding
Engines quote pages that state facts plainly. Vague pages do not get quoted.
What you do, for whom, where, and what it costs, in the HTML. Structured data that matches the page. One H1 that says the subject.
03
Trust
When two sites say the same thing, the engine names the one it can corroborate.
One consistent business name across your site, profiles and directories. Sources that mention you. A founder with a face.
The accidental block
Say no to training. Stay in the answer.
Two different kinds of crawler visit your site. One collects training data for models. The other builds the index that answers are cited from. Blocking the first is a legitimate choice. Blocking the second removes you from ChatGPT search, Perplexity and Claude. One User-agent: * rule blocks both, and nothing in your logs will tell you.
NeshatScan reads your robots.txt the way the crawlers do, line by line, and tells you which case you are in. Check yours.
Training · your choice
- GPTBot
- ClaudeBot
- CCBot
- Bytespider
- Meta-ExternalAgent
Answers · must stay open
- OAI-SearchBot
- Claude-SearchBot
- PerplexityBot
- ChatGPT-User
- Claude-User
- Perplexity-User
Sources: OpenAI · Anthropic · Perplexity · Google · Microsoft · Apple · Common Crawl · ByteDance · Meta (checked 2026-09-22)
[2]what does not matter
Half of what you will be sold does nothing. Here is the list.
A consultant who tells you what to skip is easier to trust on what to do. Each item links to the vendor's own words.
llms.txt
Optional.
A proposed convention, not a standard. No major AI company has committed to reading it. NeshatScan reports it and never grades it.
Source: llms.txt proposal (checked 2026-09-22)A composite “AI visibility score”
Not a thing.
No engine publishes a score, so any number a tool shows you is that tool's opinion. What exists are blockers you can remove and answers you can capture, dated.
Meta keywords
Ignored since 2009.
Google said so in writing sixteen years ago. Tools that still flag them are padding a report.
Source: Google — Google does not use the keywords meta tag in web ranking (2009) (checked 2026-09-22)Blocking Google-Extended to stay out of AI Overviews
Does not do that.
Google-Extended controls Gemini training and grounding. AI Overviews and AI Mode read the normal search index; the controls for those are nosnippet and max-snippet.
Source: Google — Google-Extended (checked 2026-09-22)“Submitting your site to ChatGPT”
There is no such form.
ChatGPT search cites what OAI-SearchBot can crawl. Let it in, state your facts clearly, and be corroborated elsewhere. That is the whole submission process.
Source: OpenAI — Overview of OpenAI crawlers (checked 2026-09-22)FAQ schema for rich-result stars
The stars are gone.
Google retired FAQ rich results for almost all sites in 2023. FAQ schema is still worth having for machine readability, but only for questions actually shown on the page.
Source: Google — Changes to HowTo and FAQ rich results (2023) (checked 2026-09-22)
[3]questions, answered
The questions buyers ask before they book.
[01]What is AI visibility?
AI visibility is whether AI assistants such as ChatGPT, Perplexity, Gemini and Google's AI Overviews name your business when a buyer asks who to hire, and whether they can read your site at all. It matters because an answer names a few businesses and stops: there is no second page. NeshatScan checks the reading part in seconds; its Citation Check asks the assistants a buyer's question and shows who they named.
[02]Can you guarantee ChatGPT or Google AI will mention my business?
No, and nobody honest can. What can be controlled is whether the answer engines are able to read your site at all, whether they understand what you do and where, and whether the trust signals they weigh point at you. I fix those and measure the result with dated captures, before and after.
[03]What is the difference between SEO, AEO and GEO?
SEO is being found in a list of links. AEO (answer engine optimization) is being the answer to a question. GEO (generative engine optimization) is being named inside an AI-written answer, in ChatGPT, Perplexity, Gemini or Google's AI Overviews. They share foundations, but the AI answers weigh different things: whether their crawlers are allowed in, how clearly your pages state facts, and how consistent your business looks across the web.
[04]What does NeshatScan check?
It fetches your page the way an AI crawler does and reports, with evidence: which of 14 named AI and search crawlers your robots.txt allows or blocks, indexing directives such as noindex and nosnippet, how much text a crawler that does not run JavaScript actually receives, your title, description, canonical and structured data, and whether your sitemap works. Each finding shows the exact line or value it found, and where it rests on a published rule, it links that source.
[05]Is blocking AI crawlers bad?
Blocking training crawlers such as GPTBot or ClaudeBot is your choice and NeshatScan never counts it against you. The problem is blocking the answer crawlers by accident: OAI-SearchBot, Claude-SearchBot and PerplexityBot decide whether you can be cited at all. Many sites block both with one rule and never find out.
[06]Do I need an llms.txt file?
It is optional. No major AI company has committed to reading llms.txt, so NeshatScan reports it as optional and never as a blocker.
[07]Why should I trust a consultant without a wall of client logos?
Because everything on this site can be checked, and the proof so far comes from my own products, not client work. The scan runs live on your own domain. The products I built are live: LoveJan on the App Store and Google Play, CareJan on the App Store and on the web at carejan.com, and you can scan them too. Every ranking or citation claim on this site links to a dated capture stating the query, engine and method, and there are no testimonials until a client gives written permission for one.
[08]How fast can I see results?
Access fixes take effect as soon as the crawlers return, usually days. Being named in AI answers lags behind, typically weeks to a few months, because the engines must re-crawl, re-index and re-weigh your pages. That is why the monthly plan has a three-month minimum.
[09]What do you need from me to start?
Your website address. Book a 20-minute call, bring the domain, and we start from your NeshatScan results.