TODO.md

##I still have the following things to do

• Recognize state diagram on p73 at least.
• can we implement a better way of reading text from the pdf (which is fully working using langchain py pdfdirectoryloader) that can make better assumptions on the diagram text
• is there a better way of splitting text?
• may want to deploy on streamlit for the sake of time.
• Chat history. Remember previous queries
• Work on slide deck