RAG vs. GAR: A Primer on Generation Augmented Retrieval (opens in new tab)

(blog.luk.sh)

6 pointslukol1y ago2 comments

2 comments

Echoed many of the reasons we didn't start using RAG for a recent project (Tipsiti). Curious how you ensure that users stick around while waiting for an answer? Thanks for sharing your thoughts.

lukolOP1y ago

Thanks! We've implemented a progress indicator, skeleton loaders, and object streaming using the Vercel AI SDK (see here: https://sdk.vercel.ai/docs/ai-sdk-core/generating-structured...). Especially the latter ensures that users start receiving content right away.

j / k navigate · click thread line to collapse

2 comments

lippihom1y ago

Echoed many of the reasons we didn't start using RAG for a recent project (Tipsiti). Curious how you ensure that users stick around while waiting for an answer? Thanks for sharing your thoughts.

lukolOP1y ago

j / k navigate · click thread line to collapse