psycrayLet's talk
Work

Vendor Performance Drift Detector

A weekly agent that reconciles promised ship dates against actual receipts for every active vendor, scores the trend, and alerts procurement before drift becomes a stockout.

Client
Mid-size manufacturer
Industry
Procurement
Services
AI Agents
Results

Results

  • ~85%Underperforming vendors flagged before a resulting stockout or missed shipment.
  • 4–6 hrsWeekly manual vendor review and reporting eliminated.
  • ~65%Fewer late-delivery surprises from watch-listed vendors within 90 days.

As measured during the implementation.

Agent profile

Agent profile

Watches for
  • Every week, for every active vendor
  • On-time rate below 80%
  • Three or more late deliveries in 30 days
  • A sustained downward trend
Runs
Weekly
Connects to
  • ERP purchase and receiving history
  • Vendor email
  • Google Sheets scorecard
  • Microsoft Teams
Hands off to a person
It never makes vendor decisions. The procurement lead gets the evidence and decides what happens next. Thresholds are set per vendor category, and score weights are tuned with the team’s feedback.
System

One of 3 agents in the same system

Work moves through them in order, and each agent hands the next one cleaner data.

  1. Step 1Unstructured Vendor Data NormalizerAn intake agent that reads invoices, packing slips, and shipping notices from a shared inbox, matches them to open POs, and stages clean records for the ERP’s bulk import.
  2. Step 2Vendor Performance Drift DetectorA weekly agent that reconciles promised ship dates against actual receipts for every active vendor, scores the trend, and alerts procurement before drift becomes a stockout.
  3. Step 3Supply Chain Exception EscalatorAn agent that catches short shipments, silent vendors, and buried backorder notices, checks approved alternates, and puts a ranked shortlist in front of a buyer.
Product
Vendor Drift Detector scorecard: composite scores, on-time rates, trend bars, and a Teams alert
The challenge

Years of purchase-order and receiving data sat in the ERP, never used to judge supplier performance. Promised dates were recorded but never compared with receipts at scale, so a vendor could slip from reliable to chronically late across dozens of orders before anyone noticed. Vendor reviews were anecdotal, led by whoever complained loudest.

With more than 400 vendors, weekly manual review wasn’t realistic, and the early signals in the inbox, slow confirmations and rising backorders, were never connected to purchasing decisions.

What we built

What we built

How the system works, step by step.

  1. Reconcile weekly

    Pulls PO and receiving history from the ERP and compares promised ship dates with actual receipts, line by line. On time means within two days of the vendor’s committed date.

  2. Score each vendor

    A rolling weighted score from on-time rate, average days late, 30- and 90-day late counts, and trend direction.

  3. Read the inbox

    Adds confirmation response time and backorder frequency from vendor email where the signals are consistent.

  4. Update the scorecard

    Writes results weekly to a shared vendor scorecard.

  5. Alert the lead

    Sends a Teams alert with the vendor, score, trend, and a link to its scorecard row when a vendor crosses a threshold.

Under the hood
  • ERP data
  • Google Sheets
  • Microsoft Teams
Next
AI Reviews Analysis PipelineAn analysis agent that pulls reviews from Google, Yelp, and BBB into one model, scores each by theme, and shows leadership which issues are rising, where, and why.
Contact

Have a workflow like this? Let’s map it.

Bring us the workflow that eats your team's week. We'll show you what an agent or a better system could take off its plate.

Let's talkMore work