Document Analyzer AI

Workflow Design built in stack Ai
My solution which is hosted will look like this

Inspiration

As a college student I have experienced the pain of scrolling through long PDFs, policies, or handbooks just to find one small piece of information. So I was inspired solve this problem with AI — to make documents searchable by meaning, not just keywords.

What it does

Document AnalyZer AI is an AI-powered agent that: Lets users upload files (PDFs, handbooks, notes, guides). Accepts natural language questions from the user. Analyzes and retrieves the most relevant sections.

How we built it

Document Processing: Uploaded files are broken into chunks and embedded into vector space. Vector Search: When a question is asked, the system compares it with the embeddings to find the most relevant sections. Answer Generation: An LLM (via Stack AI & Claude from anthropic) summarizes and delivers a precise answer.

Platform: Built using Stack AI for orchestration and deployment, with Claude (3.5 Sonnet) embeddings + vector database for retrieval.

Challenges we ran into

Creating a smooth user flow for uploading documents and asking questions. Handling differences between various document formats.

Accomplishments that we're proud of

I Built a fully functional hosted agent within hackathon time. Created a solution that is useful not only for hackathons but also for real-world productivity.