LaunchRepoShort answer
- There is no submission form for AI answers. ChatGPT search, Google AI Mode, Claude and Perplexity build answers from pages they have indexed or fetch on the spot.
- Step one is technical and fully yours: let the search crawlers in. For OpenAI that is OAI-SearchBot, for Anthropic Claude-SearchBot, for Perplexity PerplexityBot, for Google and Microsoft their normal search crawlers.
- Blocking training crawlers such as GPTBot, ClaudeBot or Google-Extended is a separate decision. The vendors document these switches as independent.
- Being named is the model’s choice. What helps is the same thing that helps in search: first-hand pages with facts nobody else has, and honest descriptions of your product on other sites.
- Nobody can guarantee a mention. OpenAI writes “Placement is not guaranteed”, Microsoft writes that GEO “does not guarantee grounding or citations”.
Three layers: index, retrieval, recommendation
Most advice mixes three questions. Keeping them apart shows where your effort pays off and where it cannot.
- IndexCan the engine find and store your pages at all?
- RetrievalFor this question, does the engine pull your page into the answer?
- RecommendationDoes the answer name your product as a fit?
| Layer | Question it answers | What you control | How you can tell |
|---|---|---|---|
| Index | Can the engine find and store your pages at all? | Fully: robots.txt, CDN or firewall rules, indexable pages, sitemaps, internal links. | Pages indexed in Google Search Console and Bing Webmaster Tools; the search crawlers allowed in your robots.txt. |
| Retrieval | For this question, does the engine pull your page into the answer? | Partly: pages that match the questions people really ask, with clear facts, prices and comparisons. | Your URL appears among the cited sources; Google’s Generative AI performance report in Search Console. |
| Recommendation | Does the answer name your product as a fit? | Indirectly: what your pages prove, and what independent pages, reviews and discussions say about you. | Your name in the answer text for the same prompt, checked repeatedly in fresh sessions. |
Training is a fourth, separate topic. Training crawlers collect text for future models; they do not decide whether today’s search answer cites you.
Which crawler does what
Each vendor publishes its own user-agent tokens. Search tokens decide whether you can be found for answers, user tokens fetch a page when someone asks about it, training tokens collect data for models.
| Engine | Search / answers | User-triggered fetch | Model training |
|---|---|---|---|
| Google (AI Overviews, AI Mode) | Googlebot | – | Google-Extended |
| Microsoft (Bing, Copilot) | Bingbot | – | – |
| OpenAI (ChatGPT search) | OAI-SearchBot | ChatGPT-User (may ignore robots.txt) | GPTBot |
| Anthropic (Claude) | Claude-SearchBot | Claude-User | ClaudeBot |
| Perplexity | PerplexityBot | Perplexity-User (may ignore robots.txt) | – |
Google has no separate AI crawler for Search: AI Overviews and AI Mode use Googlebot and the normal Search index. Google-Extended only covers training and grounding in some of Google’s other systems and, in Google’s words, does not impact inclusion in Google Search.
Check your robots.txt
Paste your robots.txt and a path. The check runs in your browser and applies the matching rules of RFC 9309 to each documented token: the most specific group, then the longest matching rule, with Allow winning a tie.
robots.txt is only half of it. OpenAI, Anthropic and Perplexity also ask that your host or CDN does not block their published IP ranges, so check firewall or bot-protection rules too.
What the vendors say
The statements that matter for product visibility, word for word, with what they mean for a SaaS founder.
Quotes are in the original English, as published on October 1, 2026.
| Statement | Official wording | What it means |
|---|---|---|
| Google: answers come from the index | relying on our core Search ranking systems to retrieve relevant, up-to-date web pages from our Search index | AI Overviews and AI Mode retrieve pages from the normal Search index. No index, no citation. |
| Google: eligibility | a page must be indexed and eligible to be shown in Google Search with a snippet | The technical bar is the same as for a normal search result with a snippet. |
| Google: no extra requirements | There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary. | There is no AI-only file, tag or form to fill in. |
| Google: non-commodity content | non-commodity content (such as "Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line") provides unique expert or experienced takes that go beyond common knowledge and the ordinary. | Your own tests, data and decisions count more than another summary of common knowledge. |
| Google: inauthentic mentions | seeking inauthentic "mentions" across the web isn't as helpful as it might seem. | Planting mentions of your product is not a shortcut. |
| Google: one page per query variant | doing so primarily to manipulate rankings or generative AI responses in Google Search violates Google's scaled content abuse spam policy. | Mass-producing pages for every fan-out question breaks the spam policy. |
| Google: llms.txt | Doing so will neither harm nor help your site's visibility or rankings in Google Search, as Google Search ignores them. | Fine to keep for other tools, irrelevant for Google Search. |
| Google: how to measure | use the Generative AI performance report in Search Console. | Search Console has a Generative AI performance report. |
| OpenAI: search crawler | Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though can still appear as navigational links. | Block OAI-SearchBot and ChatGPT search will not show you in answers. |
| OpenAI: ranking | ChatGPT ranks search results using multiple factors intended to help users find relevant, reliable information. Placement is not guaranteed. | Eligibility is in your hands, placement is not. |
| OpenAI: product results | Product results are selected independently by ChatGPT and are not ads, nor influenced by any OpenAI partnerships. | Product results in ChatGPT are not paid placements. |
| Anthropic: search crawler | Disabling Claude-SearchBot on your site prevents our system from indexing your content for search optimization, which may reduce your site's visibility and accuracy in user search results. | Blocking Claude-SearchBot can cost visibility in Claude’s search answers. |
| Perplexity: search crawler | PerplexityBot is designed to surface and link websites in search results on Perplexity. It is not used to crawl content for AI foundation models. | PerplexityBot is for search results, not for training. |
| Microsoft: no guarantee | SEO does not guarantee rankings or traffic, and GEO does not guarantee grounding or citations in AI experiences. | Bing and Copilot share one index; good SEO makes you eligible, nothing makes you certain. |
Six steps for a SaaS product
In order of effort. The first two take an afternoon; the rest is the work that also builds normal search traffic.
- Open the door for search crawlers
Allow Googlebot, Bingbot, OAI-SearchBot, Claude-SearchBot and PerplexityBot on the pages you want to be found with. Check bot protection on your host or CDN as well. Use the checker above.
- Get indexed in Google and Bing
Verify the site in Google Search Console and Bing Webmaster Tools, submit your sitemap and fix pages that are not indexed. Bing and Copilot share one index, and ChatGPT search sometimes queries partner search providers as well.
- Publish pages only you can write
Real tests, your own numbers, dated prices, honest comparisons with named alternatives, clear statements of what your product does not do. This is what Google calls non-commodity content.
- State the facts in plain text
Product name, category, who it is for, price and model on one clear page, consistent with your pricing page and your profiles elsewhere. Answers that compare products need these facts and pick them from somewhere.
- Be described honestly on other sites
Real listings on directories and launch platforms, review profiles with real reviews, genuine answers in communities you belong to. One honest profile is worth more than ten planted mentions.
- Measure, then repeat
Watch indexed pages and the Generative AI report, and keep a small prompt log (below). Change one thing at a time, so you can see what moved.
What you can skip
Google’s AI optimization guide lists tactics you can ignore for Google Search. Here they are with what to do instead.
| Tactic | What Google says | Do this instead |
|---|---|---|
| llms.txt or other AI files | Not used by Google Search; neither helps nor harms. | Keep one if other tools use it, then spend the time on content. |
| Chunking pages into tiny pieces | Not required; there is no ideal page length. | Structure pages for readers with clear headings. |
| Rewriting content just for AI | AI systems understand synonyms and meaning. | Write for the person with the question. |
| Seeking inauthentic mentions | Not as helpful as it might seem; spam systems apply. | Earn real listings, reviews and discussions. |
| Special structured data for AI | Not required; no special schema.org markup. | Keep normal structured data for rich results. |
The evidence on llms.txt in detail: Does llms.txt help? What it does and what is proven →
Measure without a paid panel
You do not need a paid tool to see where you stand. A fixed prompt log, filled in by hand once a month, shows changes over time.
- Write 10 to 20 questions a buyer would ask, without your brand name: “best tool for …”, “alternative to …”, “how do I …”.
- Ask each one in a fresh session with search switched on, in every engine you care about.
- Record whether your product is named, its position, and which URLs are cited.
- The cited URLs are the useful part: they show which pages the engine trusts for that question, and where an honest listing or a better page of your own is missing.
- Repeat with the same prompts after a month. Answers vary between runs, so compare patterns, not single answers.
Download the prompt log template (CSV)
We do not run or sell a model panel, and the numbers you collect this way are a sample, not a ranking. Google warns about third-party tools that claim to use internal metrics.
Where LaunchRepo stands
We apply the same rules to launchrepo.dev, and we publish where we stand rather than claim a result.
- Our robots.txt allows every documented search, user and training crawler except on /api/, where only server endpoints live (load it in the checker above).
- Our pages are indexable, listed in sitemaps and pinged through IndexNow; we also keep an llms.txt file, for agents rather than for rankings.
- In our own research on 23 September 2026 we ran generic buyer questions through a web search: LaunchRepo was not named for generic category questions and appeared for one near-brand question only through third-party directory profiles. We could not test the AI engines directly that day, so this is a proxy, not a measurement of AI answers.
- That is why we publish dated, first-hand pages such as our directory status index, tests of agent setups and comparisons with named alternatives.
- Launch directory status index →
- Directory backlinks: what we can prove, and what we cannot →
- IndexNow →
LaunchRepo helps with the part of this guide that is repetitive: honest listings on directories and launch platforms, prepared and submitted by your own agent.
Sources
Every page was read on October 1, 2026. The date column shows the page’s own last update where it shows one.
Frequently asked questions
- How do I get my product recommended by ChatGPT?
- Allow OAI-SearchBot in robots.txt and on your CDN, make sure your pages are indexed in Bing and Google, publish first-hand pages with clear product facts, and get described honestly on other sites. OpenAI states that placement is not guaranteed, so treat this as raising your chances, not buying a slot.
- Can I pay to be recommended in ChatGPT answers?
- Not in the answer itself. OpenAI states that product results are selected independently and are not ads. Ads in ChatGPT exist, but they are shown separately from the answers.
- If I block GPTBot, do I disappear from ChatGPT search?
- No. OpenAI documents GPTBot (training) and OAI-SearchBot (search) as independent settings. You can block training and still allow search. The same split exists at Anthropic (ClaudeBot vs. Claude-SearchBot) and at Google (Google-Extended vs. Googlebot).
- Do I need an llms.txt file to show up in AI search?
- Not for Google: its guide says Google Search ignores these files. Some agents and documentation tools read them. Our article on llms.txt goes through the evidence.
- How long until a robots.txt change takes effect?
- OpenAI says about 24 hours for ChatGPT search, Perplexity up to 24 hours. Google says recrawling can take from several days to several months. Being cited afterwards still depends on retrieval and ranking.
- Can a tool tell me my AI visibility score?
- Tools can sample answers, but no outside tool sees how the engines rank. Google says no third-party tool has access to its internal ranking or AI systems. Its own Generative AI report in Search Console and a simple prompt log are a solid, free start.