𝗔𝘀𝗸 𝗽𝗼𝗱𝗰𝗮𝘀𝘁𝘀 𝗾𝘂𝗲𝘀𝘁𝗶𝗼𝗻𝘀, 𝗴𝗲𝘁 𝘀𝗺𝗮𝗿𝘁𝗲𝗿

I was having a chat to Johann Van Tonder about Notebook LLM the other day. One of the joys of it is that it’s a closed loop LLM, i.e. whatever you add as the source material isn’t “polluted” by any other data (like you’d find with ChatGPT and Claude). This means that you’re only dealing with what you’ve uploaded in the workbook and you don’t get any other hallucinations or data that is out there.

A useful use case I’ve been using:
– find an area you want to learn more about
– find podcasts/panels discussions/interviews with smart people about this
– download the transcripts off YouTube and add them as workbooks to Notebook LLM
– ask questions and interrogate your source material

This gets you around all the averages you might get with a normal LLM and you’re dealing with specific people with smart opinions. Interrogating the relevant data is way easier than listening to 10 hours of opinion. You can get the combined wisdom summarised for you and probe accordingly. Asking follow up questions is helpful and you can view it as lean forward more than lean back approach
Give it a try 🙂