DocuRAG

In the world, where every partnership or cooperation based on documents, it is crucial to quickly and effectively comprehend what we are dealing with. Leverage Docusign with DocuRAG extension.

Comment

Inspiration

Signing documents can be a beginning of new opportunities or an end of success. It is scary to sign a document which you don't understand completely or miss important and hidden points. Thus, we need someone who will assist us in that simple yet risky job. DocuRAG solves this problem by extending Docusign's capability in a very different way.

What it does

DocuRAG uses RAG with AI to enable users to ask any questions about the document they are going to sign or read. Imagine your potential employer wants you to sign a work contract and performs this action using Docusign platform. The employer can use our extension app to make an appointment with our external AI web application. Employer, simply, needs to install and use the extension app in Maestro Workflow. The only prerequisite is that the extension app can be used only after "Custom Web-form" in Docusign Maestro Workflow, because it gathers required data (signer's email and google drive link to the document) from the web-form and writes these data into our system for making an appointment. The app sends a link via an email to you (signer) for accessing our web app.

How we built it

The application is considered as an extension app because it has manifest.json and does not use any Docusign APIs. It was built using "Data IO Write" extension and Google Cloud APIs. "Data IO Write" extension is responsible for writing user data into our database and Google Cloud APIs are used for authorization and RAG operations between user and LLM.

Challenges & Accomplishments

There were some challenges we faced during hackathon:

  1. Unexpected Restriction in Features of Docusign Platform.
    At first, we thought we will be able to integrate chatting mechanism into Maestro Workflow to enable users to interact with AI without being redirected to other applications. However, we overcome this obstacle by making our application "half" extension and standalone app.
  2. Cost-effective LLMs.
    It was hard to find an LLM with low-cost and effective for our project because we have already utilized many of the offered free APIs of cloud platforms. However, after long research of models and APIs, we decided to utilize Google Cloud Vertex AI as the best one in given choices.
  3. Smooth Integration of DocuRAG into Docusign.
    The last but not least (author "Snoop Dogg"), we had a hard time to enable a user to smoothly transition into our application. Making it an Integration App would be easy, but, from the start, our choice was to make this application accessible from within the Docusign platform. We cared about user experience and decided to create a "Data IO" extension app as a gateway to our web app. That way users don't have to search our application, but it will find them itself. :)

What we learned

First and foremost, we learned everything about Docusign and its developer platform. We discovered of its existence and how useful it is. Moreover, we improved our problem-solving skills by overcoming above mentioned obstacles. Google Cloud was extremely useful in our project and something we have never used before. Finally, we learned how to build a RAG system with AI and integrate it into our web app.

What's next for DocuRAG

In future releases, we think about making our app both extension and integration app to improve user experience. Also, some technical improvements will take place to make our app more robust than before.

Built With

Share this project:

Updates