Doc Inspector - AI Assistant Report Generator

Built for AI Tinkerer's Generative AI Hackathon by Michael Wiltfong and James Keezer

Deployed to https://doc-inspector-fe.vercel.app/

Status: Incomplete! The app does not yet modify the template.

This is the Front-End repo. The Back-End repo is here.

• Back-end built with Gadget - website - GitHub
• Front-end built with NextJS

Project Overview

This web app is designed to help the user extract information from a document and generate reports based on the source material. It allows users to upload a PDF, engage in AI-powered discussions about its content, and use these insights to create a new document. The new document's format is determined by a markdown file for a custom output.

Features

Document Upload and Processing

• Upload PDFs on various topics like software features or legal briefings.
• Store documents for AI processing.

AI-Powered Interaction

• Interact with an OpenAI AI assistant to get answers based on the document.
• Allow AI to generate a new document with user instructions based on this user-defined format.

User Experience Design

• A guided UX from querying information to creatively assembling a new document.

Technical Specifications

• Postgres database: Gadget.dev back-end automated DB creation, hosting, migration, and scaling.
• NodeJS: Extend backend with JavaScript functions, local or web editing.
• APIs: Gadget's GraphQL API and NPM package for CRUD, custom endpoints.
• File Storage: Gadget's secure CDN for image/media file storage with upload limits.

Considerations

• Maintain information accuracy in AI responses to prevent misinformation.
• Balance UX between querying information and creative document assembly.
• Adapt to AI technologies' capabilities and limitations.