Glance for Docusign: AI Assistant for Documents and Agreements

Project Goal: Provide instant insights on user documents and agreements in Docusign with the latest generative AI models through an intuitive AI chat interface.

App: https://glance-docusign.vercel.app/ (Beta Preview)

About the Project

glance-screenshot-airesponse

Glance for Docusign is a Generative AI Chat Assistant for your documents and agreements. Designed with an easy to use interface, Glance integrates with Docusign to provide a chatbot assistant that instantly analyzes and simplifies complex content of Envelopes, Documents, and Navigator Agreements, while offering clear insights and actionable answers.

With built-in support for over ten top-tier large language models (LLM), you can use Glance out of the box with no configuration required. We currently support the latest LLM models from Anthropic, Meta, Mistral, Amazon and Cohere.

Glance is built with security and privacy at its core, ensuring users’ private data is never stored or misused, adhering to the highest industry standards for privacy.

Key Features

Get Quick Insights on Your Agreements

  • Analyze documents in seconds with AI-powered insights tailored for agreements and contracts.

  • Receive concise and actionable summaries directly from your Docusign documents.

No Setup. Instant Support.

  • Seamlessly integrates with Docusign without the need for configurations or complex setups.

  • Ready-to-use with built-in support for Envelopes, Documents, and Navigator Agreements.

Built-in AI Models

  • Out of the box support for over ten top-tier large language models.

  • Configurable AI model settings to find the best fit for your document.

Compliant and Secure by Design

  • Ensures complete confidentiality by processing documents without storing private data.

Technical Details

Architecture Overview

glance-architecture

Web Application & API

  • The frontend web application is developed in React using the Next.js framework.

  • The backend API is built in Node.js with TypeScript powers the document parsing functionality.

  • Both service are hosted on Vercel for scalability and reliability.

Integration with Docusign

  • The application uses Docusign eSignature REST API and the new Navigator API to fetch folders, envelopes, documents and agreements.

Document parsers

  • pdf-parse: Library for converting PDF files into text.
  • mammoth: Library for converting DOCX files into text.

Security and Compliance

  • Proper authentication and handling of secrets to ensure user data remains private and secure.

  • No sensitive data is stored, including document/agreement contents, processed only in-memory for the duration of the AI analysis.

AI Capabilities

  • Custom built generative AI API on Amazon Web Services (AWS) and hosted within our own accounts, so that your data will never be used in any way including model optimisation and training.

  • Leverages state-of-the-art large language models such as Anthropic Claude, Meta and Mistral

  • Note: While the rest of the application is fully open source, the source code for the AI API is available upon request for the purpose of hackathon evaluation.

Future Plans

As we aim to provide additional insights to our users on their content in Docusign, we have plans to expand on the current offering. In particular, we are considering the following expansions:

  • Ability to use Open AI and HuggingFace

  • Saved conversation history

  • Extension app

Screenshots

glance-screenshot-welcome

glance-screenshot-docs

glance-screenshot-aicontent

glance-screenshot-aimodels

glance-screenshot-airesponse

Share this project:

Updates