Skip to content

imesut/docuexpert-hackathon

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

24 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Docuexpert

A hackathon project for Docusign.

Visit Devpost page : https://devpost.com/software/docuexpert

What is it?

Docuexpert is a connector between Docusign CLM a meeting assistants. It contains experts (agents) which also comments on agreement and keep your todos organized.

Structure

The project contains 3 applications;

  • frontend : a web app lives inside Google Docs
  • redirection-page : a simple page to store retrived accessToken for the user
  • backend : functions and data storage layer

Branches

  • main : is the public version branch
  • version : each version is merged here
  • develop : each development committed here, sinve no need to diversify feature branches.

Flow

End user uses 3 applications : Meeting Assistant, Docusign CLM (Google Docs*), and Docusign E-signature; which are shown at the top side of the diagram.

The systems utilizes, LLM (Claude), Supabase (Edge Functions, Tables and Easy Data API); whic are show at the bottom side of the diagram.

Project Diagram

  1. The systems gives an endpoint for an external meeting assistant like Fathom, to retrieve transcipts data.
  2. frontend is a SPA which lives in a Apps Script container for Google Docs.
  3. There is a bridge layer which links Google Apps Script "backend" layer to docuexpert frontend; which works in session, works with callback methods.
  4. User can set preferences and customize experts (ai agents); and take some actions;
    1. get action items from transcripts, powered by LLM
    2. apply expert's suggestions with 1-click.
    3. send the last version of the agreement to recipient.
  5. redirection page exists, because oauth tokenization cannot be whitelisted in dynamic origin urls. When a change is deployed on Apps Script the origin url might change. To to centralize token gathering, this page is prepared to hosted in a fixed origin.
  6. AI suggestions is prepared by Claude, and carried by ai-service function.

*There are no API and sandbox environment provided fot CLM, so we used Google Docs for prototyping. A good AI tool can interoperate with others and can be deployed in many forms.

Deployment

Frontend

Local Development

cd frontend
npm install
#local development
npm run dev

Production

cd frontend
#dependencies
npm install

#build & sync tool for Apps Script
npm i -D @google/clasp @types/google-apps-script

#follow oauth login to Apps Script
clasp login
#local development
npm run publish

Change the project id and other environment keys before deploy.

And build redirection-page with npm run build command, then distribute files within your website.

Backend

You can deploy per edge function to Supabase:

supabase functions deploy ai-service --project-ref <PROJECT_ID>
supabase functions deploy finalize-agreement --project-ref <PROJECT_ID>

Database - Table Schemas Database Schema

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published