Stock Balancing Recommender
Build an AI tool that compares on-hand stock against demand at every location and proposes smart transfers - moving excess from overstocked sites to short ones - so you rebalance your network instead of over-ordering.
A login-protected web tool where a planner loads location stock and demand, sees AI-proposed balancing transfers (from, to, quantity), approves only the ones worth doing, and exports a clean CSV of transfer suggestions for their WMS/ERP.
Before you start
- Per-location on-hand quantities (CSV or Google Sheet)
- Per-location demand or sales velocity (CSV or Google Sheet)
- Minimum stock levels per SKU/location, plus rough transfer cost and lead time
- Free accounts on Vercel, Supabase, and Resend (all have free tiers)
The problem this kills
You have one warehouse drowning in a SKU while another is stocking out of the exact same thing. The "easy" fix is to reorder more from your supplier - which ties up cash, fills shelves you can't afford, and ignores the stock you already own sitting in the wrong building.
Rebalancing - moving excess from overstocked sites to short ones - is almost always cheaper than buying new. But doing it by hand across dozens of locations and thousands of SKUs is a spreadsheet nightmare. You have to weigh transfer cost against holding cost, respect minimum levels at the source, and avoid double-counting the same transfer twice in a cycle. Most teams just don't, so the imbalance grows.
This Implementation Plan gives you an AI tool that does the math for the whole network and hands a planner a short, ranked list of transfers worth approving.
What you'll build
A small, login-protected web app where:
- You load per-location on-hand and demand from a CSV or Google Sheet (no integration required to start).
- The tool computes imbalances - which sites are long, which are short, and by how much.
- It proposes transfers (from site, to site, SKU, quantity) that move excess to need, weighing transfer cost against the cost of holding stock, and never pulling a source below its minimum.
- A planner reviews and approves the subset of transfers worth executing - nothing is committed automatically.
- Approved transfers export as a clean CSV in the exact columns your WMS or ERP expects.
What's inside the Implementation Plan
- A copy-paste runbook you feed to Claude Code, step by step - each step ends with a ready-to-paste prompt.
- It opens by interviewing you about your business - your locations, how your stock and demand data is shaped and named, your SKU conventions, your typical and peak volumes, and your real approval rules. The tool is then tailored to how you actually work, not built from a generic template.
- A data model and validations shaped by your answers.
- The balancing logic: imbalance scoring, transfer cost vs holding cost trade-off, minimum-level protection, and de-duplication.
- The approval screen and the CSV export in your system's column layout.
- A "No API yet?" fallback so the whole thing is buildable today from spreadsheets alone.
The governance it includes (this is the point)
This is not a script that quietly rewrites your inventory. It's built like real internal software:
- Login so only your team can use it.
- Row-level security so each organization only ever sees its own data.
- A full audit trail - who proposed, who approved, what changed, and when.
- A hard human-in-the-loop approval gate - the AI drafts transfer suggestions, a planner reviews and approves, and only approved transfers make it into the export.
- Duplicate guards so the same SKU + transfer pair can't be proposed twice in one cycle.
Who it's for
Distribution planners, supply planners, and regional managers who run stock across more than one location and are tired of reordering when the answer is already on a shelf somewhere in the network. No coding experience needed - if you can use a spreadsheet and paste text, you can build this.
You've got this - paste the first prompt and let the agent interview you.