I'm thinking of creating a new AI powered note-taking app which has features like 'auto suggest', 'search using natural language', 'asking natural language question to get answer using notes as knowledge base ', etc.
The best note-taking app for me would be like if I was paying a really smart person so that (i) I could tell them at any time of the day some info or thought I'd like to use later somehow, (ii) I could make queries of any format later about the data I entered.
The key thing is that I don't want to have to decide in advance how I want to use the information. I just want to store the information that I know it might be useful or important later.
(More concretely, I could see this being a single huge text file in which I can make natural language queries; each query generates a view of the file that corresponds to selected sentences in the file that match the query. For example, if I asked "what are the things I have to do for next month", then it would show me the sentences that correspond to things I have to do for next month. If I queried "exciting ML papers", then it would show me the sentences in which I link or talk about an ML paper, and I sounded excited about it.)
It would be cool if raw text could be parsed into structured data. Something like "Yesterday at 6pm, I took a flight from MIA to LAX airport on Delta" could be parsed into:
The key thing is that I don't want to have to decide in advance how I want to use the information. I just want to store the information that I know it might be useful or important later.
(More concretely, I could see this being a single huge text file in which I can make natural language queries; each query generates a view of the file that corresponds to selected sentences in the file that match the query. For example, if I asked "what are the things I have to do for next month", then it would show me the sentences that correspond to things I have to do for next month. If I queried "exciting ML papers", then it would show me the sentences in which I link or talk about an ML paper, and I sounded excited about it.)