boatbuilder

9 ms·

1. 1.▲
   Measuring LLM citations with server logs might not work as assumed(agentberlin.ai)

   3 points by boatbuilder 1y ago|1 comments
2. 2.▲
   by boatbuilder 1y ago

   I assumed you could measure LLM citations by watching server logs. Our experiment showed otherwise: ChatGPT and Perplexity often cite pages without fetching them, instead serving from a global cache. We ran this on Vercel/Next.js middl
3. 3.▲
   by boatbuilder 2y ago

   Have been using this for a while. Yep, stupidly simple and super super useful! Thank you
4. 4.▲
   by boatbuilder 2y ago

   The recent project from jina AI was a clone of this idea? https://jina.ai/reader/
5. 5.▲
   by boatbuilder 3y ago

   So is this like dspy, but has a very specific user flow I.e chat
6. 6.▲
   by boatbuilder 3y ago

   Nice. Not sure how many people would need a service like this to make it financially viable for you to maintain though. Also noticed a bug: not getting verify email