Inspiration

Docusort allows users to log in using their google email, and request files in a certain time frame or in a certain category. No longer will you have to look through your documents and PDFs to find files related to your prompt, we will give them to you. A problem that Heritage Square faced: A lack of a way to efficiently find the files they need to help them write and reply to those who use or give to their services. As a result they were looking for an easy way to get files that could be suited to their needs, and thus make finding files more efficient.

What it does

-Docusort allows the user to make a prompt such as "Give me accounting files from November 2020" and will return the links of the files they would like straight to them -Gives easy access to pdf files and documents that they are seaching for, with room for error, which OpenAI will try and fill the gaps of knowledge as best as it can. -A login feature using google drive, ensuring that no other user can access your files, thus providing security. -An instruction's page clarifying the usage of the chat bot thus ensuring even the users with no prior knowledge of the technicality of the project could use it with no hussle. -Associate files to the categories that we were given with a miscellaneous section. -This gives Kari more time to help the community, and less time spent in her desk sorting through files

How we built it

Challenges

-We ran into the challenge of finding the right database, due to our lack of experience in the field -We also ran into the trouble of navigating Google Drive's APIs but we were able to persevere in the end and learn the calls. -It costs money to maintain this, however, once we have set up the database and parsed through the files, OpenAI calls will go down significantly, so will Google Drive calls, thus allowing this to go on for longer

Accomplishments that we're proud of

Some of the achievements that we are proud of are that we learnt using Api calls better, learnt the usage of router doms in react,it was a day filled with learnings. Some vulnerabiltites that we werent able to rectify during the hackathon due to time constraints were as follows:

What we learned

How to implement a flask server with reactJS

What's next for DocuSort

  • Due to the short time frame we were not able to complete all of the goals as we wanted, but our basic layout allows for massive scalability
  • Stretch Goal 1: Implement AI into the prompt that allows you to ask for more abstract ideas with files, thus allowing more freedom and more suggestions to help you
  • Stretch Goal 2: Since we already read through the files we parse, we could use OpenAI to pick key words out of files, which can then be of use for problem statement 1, since the AI will have context clues about every file going into the conversation with the user
Share this project:

Updates