DocuInsight

9 minute read

🟢 DocuInsight is live! Check it out at www.docuinsight.ai or checkout our live demo HERE, which does not require you to create an account or anything!

🔴 This project post is an almost exact copy of the Devpost hackathon submission we submitted on January 27, 2025. DocuInsight was built by Mehmet Yilmaz and Dylan Eck for the 2024-2025 DocuSign Hackathon Unlocked. As of March 12, 2025, we did not win the hackathon, but we learned a lot and believe DocuInsight can provide real value to senders and signers. So we are focusing our efforts on making it into a real product that provides real value.

About

DocuInsight is a platform that simplifies legal contracts by highlighting key clauses, translating them into plain English, and integrating with DocuSign for e-signatures, plus offers access to legal experts.

DocuInsight was created to solve the Agreement Trap, where essential contractual details remain hidden behind legal jargon, causing people to sign without fully understanding their rights or obligations. Our vision is to help both sides of any agreement, signers and senders, gain full clarity before signing. Additionally, we wanted to provide a simple way for users to connect with legal professionals if they need personalized counsel.

Developed from December 2024 to January 2025 for the DocuSign Hackathon, DocuInsight integrates with DocuSign’s technology to import agreements or templates, utilizes the OpenAI API for text analysis, and offers a chat interface that can answer questions about the contract or providing sources to find lawyers. We believe bridging this knowledge gap will reduce confusion, legal disputes, and missed opportunities for everyone involved.

Inspiration

In early 2024, I co-founded a startup with a close friend. Trusting him, I signed initial company contracts without fully reviewing them. After investing significant time and expertise into developing the company’s MVPs with no pay, my friend used his additional equity to force me to either reduce my shares or be removed from the company. After finding a lawyer, I discovered that I had inadvertently signed away important rights and could not defend myself or the company effectively. Due to my once close friend’s selfish and reckless behavior, the dispute strained me personally and cost the company valuable time. This experience taught me the critical importance of understanding every detail in legal agreements to protect personal rights and ensure organizational stability.

What it does

Features and Functionality:

  1. Clear Contract Summaries: DocuInsight uses advanced AI models to analyze legal documents and generate concise summaries. These summaries highlight key clauses, unusual terms, potential risks, and major commitments, providing users with a clear and understandable overview of the contract.

  2. Interactive Chat Interface: The AI-powered chat feature allows users to “talk” with their contract. They can ask questions about specific clauses, the overall purpose of the agreement, or related legal concepts. The chatbot can also browse the web to provide well-sourced, accurate, and up-to-date information, ensuring users receive reliable answers.

  3. Lawyer Recommendations: For users seeking additional legal assistance, DocuInsight generates optimized search queries to recommend lawyers based on the user’s location and the type of contract they are reviewing. The search results include contact details and reviews for legal professionals, giving users an actionable way to connect with experts.

  4. Enhanced Signing Process: DocuInsight integrates seamlessly with DocuSign’s existing signing workflow. Users log in with their email, connect their DocuSign account to access templates, and send documents for signing. Once sent, the recipient can review the document, interact with the AI-generated insights, and sign as they would within the standard DocuSign platform.

  5. Post-Signing Access: After signing, users can revisit their document through the original link to review the signed agreement, access AI insights, and continue engaging with the chat interface for follow-up questions. This ensures signers have ongoing clarity about their agreements.

These features make DocuInsight a powerful tool for simplifying the complexities of legal contracts. By combining AI-driven insights, interactive support, and seamless integration with DocuSign, DocuInsight empowers signers to make informed decisions, assists senders in building trust in the agreements they distribute, and provides legal professionals with opportunities to connect with users seeking further assistance. Our goal is to foster transparency, trust, and accessibility in the agreement process, making legal text easier to understand and fairer for everyone.

If you want to see a clear video of us using DocuInsight and demonstrating its features clearly, check out this additional video we made of DocuInsight in action (this is NOT our demo video, just an additional resource if people want more details):

How we built it

DocuInsight consists of three main parts:

  • Database: Stores all the jobs, reports, user data, etc. It contains the project’s required data.
  • Frontend: The UI for the project, what the users see and interact with.
  • Analyzer: A backend service responsible for analyzing legal contracts and sending out emails.

For this project, a database was used. The database was PostgreSQL, and it was hosted on Supabase. Both the Frontend and Analyzer accessed the database using Supabase’s API through their SDKs. The database setup consisted of three tables: one for users, one for jobs, and one for reports. The users table is used for authentication and to keep track of accounts. The jobs table keeps track of contracts that need to be analyzed. The reports table stores all the reports on contracts generated by the Analyzer.

The Frontend was built using a frontend framework called Next.js. In this framework, we used HTML, Tailwind CSS, JavaScript, TypeScript, and React. These technologies enabled the frontend to function and display content dynamically and responsively. Additionally, we used several APIs in the frontend:

  • OpenAI’s API: To host and run the chatbot for each signing session.
  • DocuSign’s API: Specifically, the eSignature API was used to handle all the signing processes and provide easy access to users’ templates from their DocuSign accounts on DocuInsight.
  • Tavily API: A specialized search engine API designed for LLMs, allowing the chatbot to browse the web and retrieve the latest news or updates.
  • Resend API: Used for handling all email-related tasks, primarily for login processes on the frontend.
  • Supabase API: Accessed the project’s database hosted on Supabase to perform CRUD operations.

The frontend is fully hosted on Vercel, a hosting service developed by the creators of the Next.js framework.

The Analyzer was built using Python. It uses OpenAI’s API to analyze legal contracts using one of OpenAI’s most advanced models, o1-preview. It also uses the Resend API to send emails, mainly for sending a receipt or the email required for signing a contract on DocuInsight. The Analyzer supports Discord webhooks, allowing users to receive alerts if something critical breaks or occurs while the Analyzer is running. Like the rest of the project, the Analyzer used Supabase to access the project’s database.

The code for the Analyzer incorporates packages such as:

  • PyMuPDF, openpyxl, python-docx, and OpenAI’s o1-preview model: To load the content of various file types, including PDF, DOCX, DOC, JPG, and PNG.

The Analyzer is hosted on a Linode instance with at least 4 GB of RAM, running Ubuntu as the operating system.

Challenges we ran into

Building DocuInsight was full of technical and practical challenges. We had to figure out how to implement secure authentication that connected seamlessly to our database while ensuring user data remained protected. Understanding which parts of the DocuSign API to use for the functionality we wanted required a lot of trial and error. Structuring the frontend to provide a good user experience and integrating advanced features like LLM functionality into the existing DocuSign signing process were also tough to get right. Streaming real-time data from the server to the client for the chat interface was tricky, as was finding a way to recommend lawyers, given there is no easily accessible or affordable dataset for legal professionals through an API. We had to build web-browsing logic for the chatbot to minimize hallucinations and make its answers more reliable. Finally, making the project production-ready was the hardest part. It is one thing to build and run something locally for a demo, but it is a completely different challenge to deploy it on the web with high uptime and reliability so anyone can access it. A single failure in production could leave a bad impression, so ensuring stability was critical and extremely difficult.

Accomplishments that we’re proud of

We are proud of successfully deploying DocuInsight and making it accessible for anyone to use. Building a production-ready application that integrates seamlessly with DocuSign, handles real-world usage, and delivers reliable AI-powered insights was a significant achievement. It is one thing to build locally for a demo, but ensuring everything works in a live environment with high uptime was a challenge we are proud to have overcome.

What we learned

Throughout this project, we realized that building a solution like DocuInsight was far more complex than we initially anticipated. Parsing legal documents, ensuring accuracy, and balancing AI capabilities with affordability required careful thought and constant iteration. We also learned the importance of proper video production. To ensure our demo effectively communicated our project’s value, we started working on the video two weeks before the deadline, using professional film equipment to deliver a polished final product. Additionally, scaling projects like this presented unique challenges. Calling advanced AI models to analyze content is computationally expensive, so we had to carefully balance using the most intelligent models with making the solution cost-effective. Finally, we conducted basic market research by browsing the web, speaking with potential users, and consulting lawyers to ensure DocuInsight could provide real value. This process taught us that building something meaningful requires validating its impact rather than simply creating a solution for the sake of it.

What’s next for DocuInsight

We plan to expand DocuInsight by developing more advanced AI agents for deeper contract analysis, refining vectorized datasets for better insights, and creating a more comprehensive directory of lawyers to connect signers and businesses with legal professionals in their area. Future features include tools like an SEC case search to reference legal precedents. Our goal is to deliver even more value to signers, senders, and businesses while fostering trust and transparency in agreements.