Back to Projects
Redress
AI-powered agent for resolving consumer complaints globally.
Next.jsTavilyjsPDFTypescript
The Problem
Filing and resolving consumer complaints across companies and jurisdictions is fragmented and slow, with no single agent to draft, track, and push resolution.
The Solution
Full-stack Next.js 16 app with NextAuth, MongoDB, Groq for reasoning, Tavily for research, and jsPDF for generating formal complaint documents. Built for the AMD Developer Hackathon on lablab.ai.
// what I learned
Building a full agent pipeline for something as messy as consumer complaints (unstructured input, varying jurisdictions, no fixed schema) meant leaning on research (Tavily) to ground each case before generating anything, rather than letting the model improvise. Generating a formal PDF output (jsPDF) also forced the app to produce something genuinely usable, not just a chat response.
