🟣 Recovr: Reimagining FedEx DCA Management

A centralised, AI driven command centre transforming Debt Collection from manual spreadsheet chaos into a precise, compliance first recovery engine.

---

📑 Table of Contents

1. Problem Statement
2. The Solution
3. Tech Stack
4. Installation & Setup
5. Final Outcome

---

🚩 The Problem

FedEx manages thousands of overdue accounts through external Debt Collection Agencies (DCAs). Today, this process relies on:

• The "Excel Dump": Manual splitting of 50k+ rows takes days to allocate.
• The "Black Hole": Once assigned, FedEx has zero visibility into agency-customer communications.
• "Cherry Picking": Agencies focus on easy debts while difficult cases stagnate, with no data to prove it.

---

⚡ The Solution

Recovr is a unified platform designed to provide 360° visibility and automation.

1. 🧠 AI Driven Smart Allocation

Pain Point Solved: Eliminates the manual "Excel Dump" and random assignment.

• Solution: Our system ingests debt cases and uses AI ( Local LLMs ) to analyze details of the case and complexity. It automatically routes the case to the agency which is best suited for that specific case.
• Outcome: Allocation time reduced from days to minutes.

2. 🕵️‍♂️ The Event Timeline (Automated Sync)

Pain Point Solved: The "Black Hole" of communication.

• Solution: Using an n8n workflow, we intercept emails between the Customer and FedEx. These are parsed and instantly logged onto the FedEx dashboard.
• Outcome: A single source of truth. When a customer calls FedEx / DCA, the FedEx / DCA representatives have access to the entire timeline and context of the case.

---

🛠 Tech Stack

We utilized a modern, scalable stack focusing on rapid automation and local LLM inference.

---

🚀 How to Run Locally

Prerequisites

• Node.js & npm
• Python 3.9+
• n8n ( Required for automation workflows )
• LM Studio ( Required for AI logic )

1. Clone the Repository & Run the application

git clone https://github.com/Better-Than-You/Recovr.git

cd recovr/frontend
npm install && npm run dev

cd ../backend
pip3 install -r requirements.txt
python3 app.py

n8n start

2. Upload the workflow json file in n8n

Upload ./n8n/n8n-workflow.json to n8n and setup gmail credentials

3. LM Studio and AI

Go to developer tab, load your desired model and start the server.

Now, Recovr is up and running!!

🎯 Final Outcome

With Recovr, we have successfully moved the Debt Collection process from a reactive, opaque state to a proactive, data driven command center. We achieved a reduction in allocation time by ~99%, established a real time communication loop that removes the "black hole," and created a governance layer that ensures every dollar owed to FedEx is tracked, managed, and optimized for recovery.