[DEMO] generate blog posts or jupyter notebooks based on a list of questions and answers

[amirfz]

PROBLEM
our current transcript summarizer generates a summary of the talk takeaways, and separately a summary of the questions and answers. it would be interesting to to add a third artifact where it takes in a list of selected questions and answers from the talk to create a blog post, or a jupyter notebook to demo the ideas presented in the talk.

SOLUTION
currently the summarization is done using only llm calls, we can use the QA agent (eg. to access google search and add details beyond just the talk transcript) to write the first draft of the blog or jupyter notebook

ALTERNATIVES
n/a

OTHER INFO
n/a

[amirfz]

q: what questions are you referring to?

a: I am referring to the question and answers that the other pipeline you're working on would generate (from meeting transcription to question / answer pairs). I'm assuming that not all of them would be interesting to use as the base for a blog post. so I'm imagining your wip pipeline produces am .rst file with questions and answers, and then we would edit that file and remove anything that might not be interesting for the blog, or add any extra details we think should be there, and then feed that edited file as input to this new pipeline

[amirfz]

[amirfz]

the idea is to start from this script and modify it to be based on the configuration based system: https://github.com/Aggregate-Intellect/sherpa/blob/main/scripts/transcript_summarizer_aws_lambda.py

[artificialfintelligence]

Updated diagram (closer to the version that was demo'ed):