Inspiration
We have thousands of documents and messages across all our services, from Notion to Drive to Slack, and we have the hardest time finding the information we're looking for.
What it does
We can find the exact right paragraph that you're looking for across all your files, even if none of the keywords match. For example, you can say, "Find me that passage where we talk about our company's strategy towards China", and it'll pull up the perfect matching paragraph from a report you forgot about in an old Google Drive folder.
How we built it
We create embeddings of every single paragraph and message across all your services and find the nearest neighbor.
Challenges we ran into
Many of the APIs we tried to use to download a user's data were far more difficult to use than they should have been.
Figuring out how to combine or split apart passages to get the best embedding search results was also difficult.
Accomplishments that we're proud of
We've got it working on Google Drive and Slack this weekend!
What we learned
The problem of storing and retrieving closest matching paragraphs is far less trivial than we thought.
What's next for Search All Your Stuff
Add all your other services and make a frontend!
Built With
- openai-embeddings
- pinecone
- python
Log in or sign up for Devpost to join the conversation.