Choose your AI access policy
Set search discovery and model-training choices deliberately.
You do not install ChatGPT or Claude crawlers on your site. Their services fetch pages from outside. You configure access and publish useful evidence. An embedded chatbot is a separate product and does not make an AI engine recommend you.
Your implementation route
Open SEO & GEO → Tools and settings → Robots.txt Editor. Save a copy of the current file before editing. Use page SEO settings for indexing choices. Publish and read the public robots.txt; do not replace Wix's generated rules blindly.
Platform boundaries & account requirements
Use Wix's native integrations and SEO & GEO tools. Backend examples belong in a supported Wix backend module or an external service, never in a page embed.
Open Wix ↗A written business decision on search inclusion, training use and user-requested retrieval; the effective origin and edge rules.
Follow the steps
Check only after completingSeparate the provider controls
For OpenAI, review OAI-SearchBot for search and GPTBot for model development separately. For Anthropic, review Claude-SearchBot, ClaudeBot and Claude-User. User-requested retrieval has different behavior from automatic crawling; do not assume one robots rule covers every request.
Treat Google controls separately
Googlebot supports Google Search crawling. Google-Extended controls specified Gemini training/grounding uses; it is not a separate HTTP user agent and does not directly govern Google Search ranking. Also review the current Search Console generative-AI inclusion controls linked in the Engine Map.
Implement only the intended policy
Use the selected host's settings. Preserve existing path exclusions when adding specific crawler groups: a more specific group can replace the wildcard rules for that bot. If your builder offers a bundled toggle, inspect what it actually emits. Check edge bot protection too.
Publish evidence that can answer a buyer
Put service scope, location, prices/cost drivers, limitations, author expertise and supported claims in accessible pages. Link related detail. An llms.txt file can support particular tools but is not a universal admission ticket or a substitute for crawlable pages.
Implementation recipes
Expand · choose · copy</>Example policy: search allowed, training restrictedOpen recipe
Custom robots reference only. This example assumes no path exclusions. Preserve your site's real restrictions in each specific group. Provider controls differ; Google-Extended also covers specified grounding uses.
User-agent: *
Allow: /
User-agent: OAI-SearchBot
Allow: /
User-agent: GPTBot
Disallow: /
User-agent: Claude-SearchBot
Allow: /
User-agent: ClaudeBot
Disallow: /
# Review Gemini training/grounding tradeoffs before using this:
User-agent: Google-Extended
Disallow: /
Sitemap: https://example.com/sitemap.xml
# User-triggered fetchers and edge firewall policies are separate.
# Recheck current provider documentation before publishing.Save the public robots response and the provider-policy links with today's date. Confirm that the rules match the intended search/training choices. A crawler user-agent string in a log is not authentication or proof of recommendation.
Revert only the changed bot groups or dashboard toggle. Recheck the live file and edge rules. Keep a dated record of why the policy changed.
Read the primary instructions
Guidance checked October 7, 2026. “2027” describes this prototype’s visual concept, not a prediction of platform rules.