Back to Portfolio

AI Supply Chain Copilot (LLM Agent)

An advanced prototype of a Large Language Model supply chain copilot. The agent operates under the **ReAct (Reasoning + Acting) Agent Framework**, dynamically querying local databases, running safety stock algorithms, and drafting printable S&OP purchase orders.

💬 AI Supply Chain Analyst Chat
Hello! I am your AI Supply Chain Copilot. I have real-time read access to your GCC Inventory logs, vendor transits, and demand projections. Click one of the pre-set prompts below or ask me a custom question!
⚠️ Muscat Stockout Risk Check
📑 Draft PO for Oud Royale 100ml
📦 S&OP Riyadh Status Overview
🧠 Agent LLM Thought & Action Console (ReAct)
[Thought] Agent online. Awaiting system query...
🔬 Technical Architecture

This copilot operates under the **ReAct (Reasoning + Acting)** framework. Instead of simply generating text, the agent executes reasoning steps to fetch parameters and invoke specialized tools.

DB Query Tool: Queries an in-memory SQL database holding GCC store records, stock counts, and daily replenishment rates.

PO Generator: An automated document compilation tool that calculates exact order sizes using safety stock models and builds structural invoices.

Vinayak Bhadani — Demand planning & S&OP in Dubai, building supply chain tooling for GCC operators. Every model here is public: the code and commit history are on GitHub.