WMS AI Copilot dashboard and chat interface

WMS AI Copilot - Warehouse Management AI Assistant

Next.jsTypeScriptExpressGroq Cloud (Llama 3.x)Vercel AI SDKSupabase (PostgreSQL)Tailwind CSSshadcn/uiTanStack Query & Tablenext-intlJestnpm Workspaces

Full-stack, real-time AI copilot for warehouse management built with Next.js, Express, Supabase, and the Vercel AI SDK, featuring streaming natural-language queries and an adaptive dashboard.

Architected and built a full-stack, production-shaped monorepo for a real-time AI copilot designed for modern warehouse management. Built for my portfolio instead of a formal job position, the platform lets warehouse admins run natural-language queries for stock levels, top sellers, and reorder alerts, returning token-by-token streaming answers or sortable data grids. Integrated Vercel AI SDK and Groq Cloud (Llama 3.x) for low-latency, deterministic tool-calling against curated, strictly typed database tools rather than raw query execution, keeping the LLM inside tight security and trust boundaries. Built a full operational split-view dashboard in Next.js App Router for real-time inventory, product, and order history tracking, and architected the app from the ground up with next-intl for fully bilingual (English/Spanish) conversational and UI flows. Structured as an npm workspaces monorepo with strict TypeScript typing across API, frontend, and AI tool definitions, backed by Jest test coverage.

Visit live project

Features

Adaptive Visualization Pipeline

An intelligent frontend engine dynamically renders plain text, charts, or sortable TanStack tables based on the schema of the AI's query result — shown here answering a natural-language request with a live pie chart and category breakdown.

Sortable Data Grids

For tabular results, the copilot streams structured data straight into a sortable TanStack Table, letting admins reorder and scan inventory results without leaving the conversation.