Redress

Most people have a legitimate complaint against a bank, telco, utility provider, landlord, or government agency but have no idea how to formally escalate it. They either give up, send an ineffective email, or don't know which regulatory body to contact. Redress solves that. Redress is a full stack web application powered by an AI agent that guides users through a three-stage complaint resolution flow. In the first stage, the agent understands the complaint by asking up to three targeted clarifying questions to gather the key details. In the second stage, it generates a professionally structured formal complaint letter with the correct recipient, recommended sending channel, and relevant regulatory body details for that country and sector. In the third stage, if the user indicates their complaint was ignored, the agent generates an escalation letter addressed directly to the regulatory body, including step-by-step filing instructions. Every letter is displayed on screen and available as a downloadable PDF. Users can manage multiple complaints, view their full history, and track the stage of each case. The agent uses Tavily web search to fetch live regulatory body contact details, making it accurate for any country and any sector including banking, telecom, utilities, housing, government services, and e-commerce. Built with Next.js 16 App Router, MongoDB, NextAuth with Google OAuth, Zustand, Tavily, and jsPDF. The agent logic lives entirely in server-side Next.js API routes calling the LLM, with the AMD Developer Cloud API as the intended provider. Note: AMD Developer Cloud credits were applied for on May 5th but were not received before the submission deadline despite following all required steps. The agent is fully compatible with AMD's OpenAI-compatible API and can be switched by updating two environment variables.