Suggestion from Cyberinfrastructure Alignment workshop participant: "Could incorporate an LLM (with RAG) to allow basic questions across all the different site-hosted guides/pages. LLM would be trained on the site, to ensure that information is only coming from a trusted source AND would direct users to relevant pages/guides." And comment from another participant: "I just made one of these for production at my job (BI/BA algorithm use case) if you need code (it is hybrid and blends open source with the OpenAI API), it only takes about .03 per call."
Suggestion from Cyberinfrastructure Alignment workshop participant: "Could incorporate an LLM (with RAG) to allow basic questions across all the different site-hosted guides/pages. LLM would be trained on the site, to ensure that information is only coming from a trusted source AND would direct users to relevant pages/guides." And comment from another participant: "I just made one of these for production at my job (BI/BA algorithm use case) if you need code (it is hybrid and blends open source with the OpenAI API), it only takes about .03 per call."