I ask because apparently CloudFlare has launched theirs: https://developers.cloudflare.com/llms.txt<p>Has anyone seen it work affect their AI results?<p>I know it's still in the proposal stage: https://llmstxt.org/<p>and many here dont like it for various reasons:
https://news.ycombinator.com/item?id=41439983<p>But what evidence is there now that it is being used as a data source over the actual content of the website?
I searched Google using "inurl:/llms.txt" and mostly found blogs talking about it but there are a few real world cases apparently. There is probably a better way to search for it. My search skills are awful. If there was a way to add a limit to type "text/plain" that might weed out the blogs.
I think it's still early stage, and it's likely LLMs running search will add a "checkForLLMsTxt" step to their pipeline. But as models get smarter, it's unclear how much value you'd get from the LLMs.txt vs just having the model read your website.<p>We had it set up for our API docs[1] (since it's provided by Mintlify). And it's nice to know it's there. But unclear how much traffic those LLMs.txt pages get.<p>[1]<a href="https://docs.getomni.ai/llms.txt">https://docs.getomni.ai/llms.txt</a>