Fesoni

• 

  additional style template

Inspiration

Shopping online often feels overwhelming — endless scrolling, too many options, and no guarantee that products fit your personal style. We wanted to solve this problem by combining AI's ability to understand aesthetics with automated document generation to create something more personal, efficient, and inspiring: a digital shopping assistant that actually gets your vibe.

What it does

Fesoni is a web app with a conversational interface where users describe their style preferences or lifestyle needs. The AI asks clarifying questions, interprets their vibe, and fetches curated Amazon products in real-time. The magic happens when Foxit APIs transform those results into polished, personalized style documentation — from lookbooks and seasonal guides to quick-buy style cards — optimized for mobile, social, and print.

How we built it

• OpenAI Integration: Built smart conversation flows to interpret abstract aesthetics into concrete shopping attributes.
• RapidAPI Amazon: Integrated Amazon product search for real-time product discovery.
• Foxit Document Generation API: Automated the creation of personalized style portfolios and guides.
• Foxit PDF Services API: Enhanced outputs with compression, splitting, merging, and conversion for multiple formats.
• Frontend: Developed a React-based web app with a simple chat-style interface.
• Backend: Built with Node to orchestrate data flow between AI, Amazon, and Foxit APIs.

Foxit Integration Overview

Our implementation leverages Foxit's dual-API architecture to fulfill the core requirement of dynamically generating documents from structured data and applying multiple PDF enhancements. Here's how each API was utilized:

Document Generation API Integration

• Template Analysis (/api/AnalyzeDocumentBase64): Debugs Word templates to identify available merge fields for dynamic content insertion
• Document Generation (/api/GenerateDocumentBase64): Creates personalized PDFs and DOCX files from Word templates, merging structured style data (user preferences, curated products, styling tips, color palettes) with template placeholders
• Hyperlink Processing: Converts product URLs into clickable links within generated documents using Word's hyperlink format

PDF Services API Integration

• Document Upload (/api/documents/upload): Uploads generated documents to Foxit's processing system, receiving document IDs for subsequent operations
• Compression (/api/documents/modify/pdf-compress): Applies configurable compression levels (HIGH/MEDIUM/LOW) to reduce file sizes for mobile and web sharing
• Document Splitting (/api/documents/modify/pdf-split): Creates focused quick-reference cards by extracting specific pages from comprehensive style guides
• Content Extraction (/api/documents/modify/pdf-extract): Extracts text, images, or specific pages for repurposing content across different formats
• Image Conversion (/api/documents/convert/pdf-to-image): Converts PDF pages to high-resolution images (customizable DPI) for social media sharing
• Document Combining (/api/documents/enhance/pdf-combine): Merges multiple style documents with bookmarking and table of contents generation

Workflow Implementation

Our system implements the required workflow through a chained process:

1. Dynamic Generation: Transforms user style data into personalized documents using Word templates
2. Parallel Enhancement: Simultaneously applies multiple PDF services (compression for mobile optimization, image conversion for social sharing, page extraction for quick-reference cards)
3. Async Task Management: Monitors processing status and handles document downloads through Foxit's task management system

The document generation integrates status tracking for all API operations.

Challenges we ran into

• Translating vague style descriptions (like "dark academia" or "cottagecore") into structured attributes for product search.
• Balancing AI-driven personalization with real-time Amazon data.
• Designing a chained workflow with Foxit APIs for multi-format delivery without performance bottlenecks.
• Managing document generation speed while keeping visual quality.
• Template debugging: Ensuring Word template merge fields aligned with our dynamic data structure using Foxit's template analysis tools.

Accomplishments that we're proud of

• Built a system where a user can type "I need something for job interviews" and instantly receive a curated interview-ready style guide PDF.
• Successfully implemented a chained Foxit workflow to generate optimized, mobile-ready style portfolios.
• Created a flexible AI questioning system that adapts to both specific shoppers and exploratory users.
• Multi-format document delivery: Users receive comprehensive PDFs, social-ready images, quick-reference cards, and editable DOCX files from a single interaction.

What we learned

• How to combine AI-driven aesthetic understanding with structured product search.
• The power of Foxit APIs in taking raw product data and turning it into shareable, polished deliverables.
• The importance of adaptive user flows for handling both clear intent and uncertain preferences.
• Document workflow optimization: Balancing generation speed with quality through parallel processing and strategic use of Foxit's enhancement APIs.

What's next for Fesoni

• Expand to multi-retailer integrations beyond Amazon and also brands APIs.
• Add community features (users can remix style guides).
• Introduce AI-driven wardrobe planning over time (tracking past guides to suggest future purchases).
• Mobile app version with AR try-on capabilities.
• Advanced Foxit integration: Implement document merging for seasonal lookbook collections and PDF form generation for size/fit tracking.

Built With

• ai/ml
• foxit-api
• git
• javascript
• node.js
• openai
• react
• tailwind-css
• typescript