Manufacturer — production reporting automation

The problem

Production data lived in ERP; quality checks in a separate system; scrap and rework tracked in Excel on the shop floor. The ops manager re-keyed adjustments every Friday. Leadership asked for a dashboard; IT quoted six months. Plant managers still called for numbers that should be self-serve.

Constraints

• — ERP is system of record — no changes to shop-floor OT network
• — Reports needed by 8am Monday — no manual export step
• — Plant teams are not technical — output must be plain English summaries
• — Fixed project budget — not an open-ended BI programme

Approach

1. Step 1

   Workflow map & ROI

   Mapped the Monday reporting loop — time per step, error rate, and who owned each export. Quantified 8+ hours/week of manual reconciliation to build the business case.

2. Step 2

   Integration pipeline

   Built Power Automate and Logic Apps pipeline — scheduled ERP and quality exports, normalised data in Azure Data Lake, and refreshed a Power BI dataset overnight.

3. Step 3

   AI executive summaries

   Azure OpenAI generates a plain-English Monday narrative — production vs plan, quality exceptions, inventory alerts — with human review before email distribution to leadership.

4. Step 4

   Monitoring & handover

   Teams alerts on pipeline failures, runbook for replay and data corrections, and training for ops manager to adjust distribution lists.

Deliverables

• Automated ERP-to-reporting pipeline — nightly refresh
• Power BI dashboard — production, quality, and inventory KPIs
• AI-generated Monday executive summary with review gate
• Failure alerting and replay runbook
• Process documentation for ops and IT handover

Results

Reporting time

Monday pack prep dropped from ~8 hrs to under 30 min review

Data freshness

Dashboard live by 6am — no manual exports

Accuracy

Single source of truth — eliminated re-keying errors

Stack

Power Automate · Logic Apps · Azure Data Lake · Power BI · Azure OpenAI