- Utilize LangChain for constructing GPT-powered chatbots equipped with tailored knowledge of a given dataset (by scrapping website into txt file).
- Employ text embeddings alongside a vector database to execute retrieval-augmented generation (RAG) seamlessly.
(reference: https://www.youtube.com/watch?v=HSZ_uaif57o&t=5323s&ab_channel=freeCodeCamp.org)
-
how can I make font adjustable (shrink size when I do ctrl + -)
-
play with CSS
-
try it with the law society website?
-
How to make it embedable?
-
Improve the document upload process give it a UI
-
import and chunk PDF into vector DB
-
authentication + rate limit
-
use firebase instead
-
reference
-
cost estimation
-
run on local model
-
save chat history and analysis ( sentiment )
-
start with an AI conversation first ( offer a few example prompts )
-
how to reduce cost? (record answers to questions)