InsightFlow: AI Supply Chain Sentinel

Inspiration ​Supply chains are drowning in paperwork and manual checks. I wanted to build a tool that allows a manager to simply "show" the AI a document or a photo and get instant analysis, reducing hours of manual data entry to seconds. ​⚙️ How it works ​Document Intelligence: Uses Gemini’s large context window to compare supplier PDFs against company policy. ​Visual Audit: Uses Gemini Vision to analyze photos of inventory or labels to ensure compliance. ​Natural Language Query: A chat interface built with React where users ask, "Which of my suppliers are at risk based on the latest news?" ​🛠️ Challenges we ran into ​Mapping unstructured data from different document formats (PDFs, JPGs) into a structured MongoDB database was tricky, but we solved it by using Gemini's structured output capabilities. Gemini Integration ​You can copy-paste this into your submission! ​"Our application, InsightFlow AI, leverages the core strengths of Gemini 3 to provide a seamless multimodal experience. Specifically, we utilize Gemini’s Reasoning capabilities to cross-reference data between uploaded images (like shipping labels) and text-based contracts. ​The Large Context Window is central to our app, allowing it to ingest lengthy 50-page supply chain audits and extract key risk factors without losing context. We also integrated Gemini’s Multimodal capabilities, enabling the app to 'see' and 'read' simultaneously. This is not just a chatbot; it is an intelligent layer that sits between raw logistics data and decision-makers. By using the Gemini API, we’ve eliminated the need for separate OCR and NLP pipelines, creating a faster, more accurate, and unified solution for global trade monitoring." ​🚀 Your Next Steps (To-Do List) ​Frontend: Build a simple React page with a "File Upload" button (for Images/PDFs). ​Backend: Create a Node.js route that sends these files to the Gemini API. ​Video: Record a 2-minute demo. Even if the UI is simple, show that the AI can "read" the file you uploaded.

Built With

• css3
• gemini
• mongodb
• node.js
• python
• react.js