Be the answer AI gives your buyers.
ExpertLayer is a generative engine optimization firm. We build microsites that ChatGPT, Claude, Perplexity, Gemini, Grok and Google AI Overviews read, name and cite. Then we track every crawl, mention and citation, so you can see it working.
A buyer asks
Which freight audit software fits a mid sized shipper?
The engine answers
It usually comes down to three vendors. YourBrand is the common first pick for mid sized shippers, with published pricing and a two week setup.
1your-microsite.com/compare
Behind that answer, the crawlers that read the page
- 09:14:02OAI-SearchBot200/compare/yourbrand-vs-vendor-b
- 09:14:09ClaudeBot200/pricing
- 09:14:16PerplexityBot200/best-freight-audit-software
- 09:14:23ChatGPT-User200/compare/yourbrand-vs-vendor-b
- 09:14:30Googlebot200/sitemap.xml
- Search
- On request
- Training
The six surfaces we work on, and the crawlers that feed each one
-
ChatGPT
- OAI-SearchBot
- ChatGPT-User
-
Claude
- Claude-SearchBot
- Claude-User
-
Perplexity
- PerplexityBot
- Perplexity-User
-
Gemini
- Googlebot
- Google-Extended
-
Grok
- No documented crawler
-
Google AI Overviews
- Googlebot
An AI answer is built from pages a crawler could read.
Before an engine can name you, four things have to happen, in this order. Most company sites fail at the first one.
-
Step 1
A crawler fetches the page
Most AI crawlers do not run JavaScript. If your content appears only after scripts load, they receive an empty shell and move on.
-
Step 2
The engine keeps what it understood
A plain question, a direct answer, a table and structured data survive. Clever copy and buried facts do not.
-
Step 3
The answer names a source
A mention puts your name in the answer. A citation puts your link under it. We count both, one question at a time.
-
Step 4
The buyer clicks through
Now it is a visit to a page you own, from someone who asked a buying question a minute ago.
Know who is at the door.
Every AI company sends more than one crawler, and each one has a different job. We read the logs by role, because a visit from a training crawler and a visit from a search crawler mean very different things for your pipeline.
Three roles do the work, plus a control token that is not a crawler at all. Only search crawlers and on request fetchers can put your link under an answer.
OpenAI
Feeds ChatGPT
- OAI-SearchBotSearchIndexes pages so ChatGPT search can show and cite them.
- ChatGPT-UserOn requestVisits a page when a ChatGPT user or a custom GPT asks for it.
- GPTBotTrainingCollects pages that may be used to train OpenAI models.
Anthropic
Feeds Claude
- Claude-SearchBotSearchIndexes pages to improve the search results Claude draws on.
- Claude-UserOn requestVisits a page when a Claude user asks a question that needs it.
- ClaudeBotTrainingCollects pages that may be used to train Claude models.
Perplexity
Feeds Perplexity
- PerplexityBotSearchIndexes pages so Perplexity can surface and link them in answers.
- Perplexity-UserOn requestVisits a page when a Perplexity user asks about it.
Feeds Gemini, AI Overviews and AI Mode
- GooglebotSearchBuilds the Google index. AI Overviews and AI Mode answer from that same index.
- Google-ExtendedControl tokenDecides whether your pages may be used for Gemini training and grounding. It does not affect Google Search.
xAI
Feeds Grok
xAI publishes no crawler documentation. There is no user agent list, no published address range and no stated robots.txt policy. Names reported in logs, such as GrokBot and xAI-SearchBot, are unverified, so we treat a Grok visit as unconfirmed until the source address checks out.
Microsoft
Feeds Copilot
- BingbotSearchBuilds the Bing index, which Microsoft Copilot answers from.
This is the short list. Read the full AI crawler field guide, with every company and the robots.txt rules that go with it.
We build the site the engines want to quote.
A microsite is a focused site on its own domain, built around the questions your buyers type into AI. It has one job: be the clearest, most current and most quotable source on your category. It goes live fast, because nothing on your main site has to change.
your-microsite.com/freight-audit-software-for-mid-sized-shippers
Which freight audit software fits a mid sized shipper?
It depends on load volume and carrier mix. Below 5,000 loads a month, the choice usually comes down to three vendors. The table compares them on price, setup time and integrations.
| Vendor | Pricing | Setup |
|---|---|---|
| YourBrand | Published | 2 weeks |
| Vendor B | On request | 6 weeks |
| Vendor C | On request | 10 weeks |
Published August 4, 2026. Updated November 12, 2026.
{ "@type": "FAQPage", "mainEntity": [ ... ] }
Served as finished HTML to OAI-SearchBot, ClaudeBot, PerplexityBot and Googlebot.
-
A
The question is the headline
The page title is the question a buyer asks, in the buyer's own words.
-
B
A direct answer comes first
Two or three sentences an engine can lift whole, before any detail.
-
C
Facts sit in a table
Prices, setup times and fit in rows and columns, not buried in paragraphs.
-
D
Every page carries its dates
Published and updated dates on the page, in the sitemap and in the markup, and all three agree.
-
E
Structured data says what the page is
JSON-LD tells a machine whether it is reading an article, a FAQ, a comparison or a list.
-
F
It arrives as finished HTML
Prerendered before the crawler asks, so a bot that does not run JavaScript still gets every word.
What goes on a microsite
- Answer pages
- One buyer question, answered at the top and backed up below.
- Comparison pages
- Two named options, side by side, on the criteria buyers use.
- Alternatives pages
- Who else to consider, and who each option fits.
- Pricing pages
- What things cost, in public, with the date it was checked.
- Directories
- The whole category in one place, graded on the same scale.
- Change logs
- What changed and when, which is how a site proves it is current.
Seven tools. One loop.
This is the stack we work in every day. Each layer does one job and hands its result to the next, and what the last layers show us decides what the first one builds tomorrow.
-
1
Base44
Build
Where the microsite is built. Pages, data and structured markup live in one app, so a new comparison page or a corrected fact can ship the same day.
Hands off to the next layer. A live site, ready to be read.
-
2
Hado
Serve and log
Sits in front of the site at the DNS level. Every crawler receives a finished HTML page instead of an empty JavaScript shell, and every bot visit is logged by crawler name.
Hands off to the next layer. Readable pages, and a record of which bots came.
-
3
Google Search Console
Index
Confirms what Google has indexed and shows the searches each page appears for. Google AI Overviews answer from that same index, so this is where a new page first proves it exists.
Hands off to the next layer. Indexed pages, and the queries they show up for.
-
4
Otterly
Mentions and citations
Asks the AI engines a fixed set of buyer questions on a schedule and records who gets named and which pages get cited. It covers ChatGPT, Perplexity, Gemini, Copilot, Claude, Google AI Overviews and AI Mode.
Hands off to the next layer. A scoreboard with one row per question.
-
5
Google Analytics
Visitors
Counts the people who arrive and separates visits sent by AI engines from search and direct traffic.
Hands off to the next layer. How many people came, and from which engine.
-
6
Microsoft Clarity
Behavior
Shows what those visitors do on the page through session recordings and heatmaps. That is how we tell a page that gets cited from a page that gets read.
Hands off to the next layer. What to fix on the page itself.
-
7
Apollo
Outbound
Finds the companies and contacts that match your buyer, for the outbound that sends the right people to the microsite.
Hands off to the next layer. Named accounts with a reason to visit.
Then it repeats. What Otterly and Search Console show this week decides what gets built in Base44 next.
These are tools we use. The names belong to their owners, and naming them here does not imply a partnership, a customer relationship or an endorsement in either direction.
Reported one question at a time, not as one flattering number.
A single coverage score hides where you are losing. Every week you get the buyer questions that matter, one row each: where you are named, where it is close, and where you are missing.
| Buyer question | ChatGPT | Claude | Perplexity | Gemini | AI Overviews | This week |
|---|---|---|---|---|---|---|
| best freight audit software for mid sized shippers | Cited | Named | Cited | Named | Cited | Holding |
| freight audit software pricing | Cited | Cited | Cited | Absent | Named | Holding |
| alternatives to vendor b | Named | Named | Cited | Absent | Absent | Gained this week |
| yourbrand vs vendor b | Named | Absent | Cited | Named | Absent | Contested |
| how to audit freight invoices automatically | Absent | Absent | Absent | Absent | Absent | Dark. A page for it ships this week. |
Every question gets a decision
If you are cited, we defend it. If it is close, we improve the page. If you are missing, we build the page that answers it, or we stop tracking the question. Nothing sits on the report just to be counted.
Also in the weekly note
- Which crawlers visited, and which pages they read
- Which pages were cited, and for which question
- What visitors sent by AI engines did when they arrived
- What we are building next, and why
Then we send the right people to it.
Citations reach the buyers who are already asking AI. Outbound reaches the ones who have the problem and have not asked yet.
We find the accounts that match your buyer, write to each one about something we can see from the outside, and point them to the page on your microsite that answers their question. Every email has somewhere useful to send the reader, and every visit shows up in the same reporting.
Questions people ask first
The longer list, including the outbound service, is on the FAQ page.
What is generative engine optimization?+
Generative engine optimization, or GEO, is the work of getting a company named and cited in the answers AI engines write. That means ChatGPT, Claude, Perplexity, Gemini, Grok and Google AI Overviews. It covers three things: making pages readable to AI crawlers, writing them so an engine can quote them, and measuring mentions and citations one buyer question at a time.
How is GEO different from SEO?+
SEO earns a position on a results page. GEO earns a place inside the answer. The groundwork is shared, because a page still has to be crawled and indexed before anything can happen. The difference is the unit of success. In SEO it is a ranking for a keyword. In GEO it is a mention or a citation for a question a buyer asked.
Why build a microsite instead of fixing our main site?+
Speed and focus. A microsite is built for crawlers from the first line, covers one category in depth, and goes live without touching your main site, its content system or its release calendar. What proves out on the microsite can be carried back to the main site later.
Which AI engines do you work on?+
ChatGPT, Claude, Perplexity, Gemini, Grok, Microsoft Copilot, and Google AI Overviews and AI Mode. We track mentions and citations on the engines our measurement tool covers, and we read the crawler logs for every bot that identifies itself.
Can you guarantee that AI will cite us?+
No, and nobody honestly can. The engines decide what they cite. What we control is whether your pages can be read, whether they answer the question better than what is cited today, and how quickly we find out when something changes.
How soon will we see something?+
Crawler visits come first, often within days of launch. Indexing follows. Mentions and citations come after that, and they move one question at a time rather than all at once. We report each stage as it happens so you are never waiting on a single number.
What does the outbound part add?+
Citations reach the buyers who are already asking AI. Outbound reaches the ones who have the problem and have not asked yet. We find the accounts that match your buyer, write to each one about something we can see from the outside, and send them to the page on your microsite that answers their question.
What do you need from us to start?+
Your website, your category, the competitors you lose to, and the questions your buyers ask before they call you. We do the rest, starting with a free visibility read.
See what AI says about your category.
Send us your site and who you compete with. We come back with a visibility read: a short document that shows where you stand today. It is free, and there is no call to sit through.
- Ten questions your buyers ask, run across the major AI engines
- Who gets named in each answer, and which pages get cited
- Where your company appears today, and where a competitor does instead
- Whether the AI crawlers can read your site at all
- The first three pages we would build