AI Inventory & Operations Dashboard
Full-stack inventory workflow concept with AI reorder suggestions in development
Full-stack inventory dashboard in development — architecture, schema, and role-based access are built; OpenAI forecasting is in prompt testing, and the inventory workflows come next.
Problem
Small-to-mid-size businesses rely on spreadsheets and gut instinct for inventory decisions, leading to stockouts or costly overstock.
Goal
Build a modern dashboard that combines real-time inventory tracking with LLM-powered forecasting to surface actionable restocking recommendations.
What I Built
A Next.js full-stack app with role-based access and a Prisma-backed PostgreSQL data layer, designed around OpenAI demand analysis that will generate plain-language reorder suggestions — structured-output prompts are currently in testing.
Key Features
- Role-based access model (admin, warehouse, viewer)
- Prisma-backed PostgreSQL schema
- Planned: AI demand forecasting via OpenAI structured outputs (prompts in testing)
- Planned: Real-time inventory grid with search, filters, and bulk actions
- Planned: Automated low-stock alerts with configurable thresholds
- Planned: CSV import/export for legacy system compatibility
Challenges
- Designing structured prompts that return reliable JSON for forecasting
- Planning real-time updates that stay efficient on the database
- Building a permission model flexible enough for multiple org sizes
Outcome
Architecture, schema design, and role-based access are complete. OpenAI integration is the active workstream — demand forecasting prompts are in structured-output testing.
Current Status
Next phase: ship OpenAI forecasting, then build the planned inventory workflows.