Procurement Verified Real deployment at a named or precisely described company, with a measured result from a primary source.

AI replenishment recommendations so reorders are not missed

Added 2026-10-10 · Last updated 2026-10-14

Role

Purchasing · Typical users: purchasing manager, buyer, materials manager

The problem

Reorders are decided from spreadsheets or static min/max settings. With thousands of raw materials and long, unstable lead times, buyers miss items that are about to run out and overbuy others, so production stops for a missing part while cash sits in excess stock. AAI (US fire-sensor maker) managed about 3,000 raw materials and 550 products on spreadsheets, with frequent stock-outs; GRPI (US inks and coatings maker) had a three-month lead time on raw materials and planned in spreadsheets with little visibility across thousands of SKUs.

Step by step

Clean the item master first: unit of measure, supplier, lead time, MOQ and order multiple for every purchased item. GRPI says clean ERP data was what made the tool work.
Export 12–24 months of usage (issues to production or sales) per item, current on-hand, open POs and open orders.
Classify items by value and usage (ABC/XYZ) so the critical and fast-moving items get the most attention.
Calculate safety stock and reorder point per item from usage variability and the real lead time (from PO date to receipt, not the supplier's catalogue figure). Do this in the planning tool or a spreadsheet, not by asking a chatbot to do the arithmetic.
Each day, list items whose projected stock (on-hand + on order − demand over the lead time) falls below the reorder point, with a suggested quantity rounded to MOQ and order multiple.
The buyer reviews the list, checks it against known changes (new product, lost customer, supplier shutdown), groups lines by supplier and releases the POs in the ERP.
Review weekly: stock-outs, fill rate, excess and slow-moving stock, and the items where the buyer overrode the suggestion and why.
Raise safety stock deliberately where lead times are long or unstable, as Netstock recommended at AAI, and record the reason.

Tools

Netstock integrated with Sage 100cloud: used at AAI
Netstock Opportunity Engine integrated with Fishbowl: used at GRPI
Other tools to evaluate: Streamline (GMDH), Prediko, Inventory Planner, and the MRP / replenishment modules of your ERP
Do-it-yourself: a spreadsheet or Python script for the reorder-point maths, plus an LLM to draft the buyer's summary and supplier emails

Result

AAI (Sage 100cloud ERP, Netstock): daily order recommendations and item classification by value and velocity; the company reports fewer stock-outs and a 99% fill rate in its Residential division; Netstock also recommended higher safety stock where lead times exceed 52 weeks (Netstock case study). GRPI (Fishbowl ERP, Netstock Opportunity Engine, an AI feature that surfaces daily replenishment and purchasing opportunities): found about $1.5 million of excess inventory it could not see in spreadsheets, and reports reduced stock-outs and better control of raw-material planning and safety stock; no stock-out figure is given (Netstock case study). Both are vendor-published. The AI here is forecasting and recommendation; the purchasing decision and the PO stay with the buyer.

As reported by the source; most figures are vendor or self-reported and not independently verified.

Human review: cautions

Recommendations are only as good as the lead times and usage in the ERP: wrong units of measure or lead times give confident but wrong orders.
Do not let an LLM compute reorder points or quantities; use the planning tool or a spreadsheet formula you can check, and spot-check 10 items by hand.
A buyer approves every PO; set a value limit above which a second person signs off.
Forecasts miss one-off events (a new contract, a recall, an engineering change). Planners must add known future demand manually.
Results are published by the vendor; AAI and GRPI do not give before/after stock-out counts.

Try it yourself

You are a purchasing assistant at a manufacturer. I will paste a table that ALREADY contains, per item: item, description, supplier, on-hand, on order, reorder point, suggested order quantity (calculated in our spreadsheet or planning tool), lead time in days, MOQ, and last 3 months' usage.

Tasks:
1. Do not recalculate the quantities. Check them for obvious problems: suggested quantity below MOQ, not a multiple of the order multiple, more than [6] months of usage, or an item with zero usage being reordered. List each problem.
2. Group the items to order by supplier and draft one PO summary per supplier (lines, quantities, requested dates = today + lead time).
3. Draft a short email to each supplier asking them to confirm price and delivery date.
4. List items where I should check with planning first (big jump in usage, new item, very long lead time).
Do not send, order, post or change anything in any system: produce drafts for my review and approval.

Table:
[paste]

Want more like this? Join the newsletter

Related cases

Agent sets order quantities and inventory from the production plan