CASE STUDY
€100M Revenue Hospital
9 DCs
340 Suppliers
CASE STUDY
€100M Revenue Hospital
9 DCs
340 Suppliers
Revenue: €100M | 300 employees | hospital ops
Ops: consult notes, structured file, medical letter
Tech Stack: existing clinical workflow
Supply chain exception management was entirely manual. When a disruption hit, the ops team had to check SAP for inventory, Manhattan for warehouse allocation, Blue Yonder for demand forecast, then call or email 3-4 people to coordinate the response. Average disruption response time was 2-4 days.
Stockout rate was 6.8%. The ops team ran a daily war room where 12 people spent 90 minutes going through an Excel tracker, mostly status updates. Supplier performance data existed across 3 systems with no unified view. The ops team was the integration layer between these systems. When someone quit, their institutional knowledge walked out with them.
Demand planning ran a monthly S&OP cycle that was already outdated by publication. Warehouse allocation was a nightly batch process. MRP-generated POs had a 23% amendment rate. Inbound logistics was tracked in a spreadsheet by 4 people. Quality data was disconnected from supplier scorecards. Returns processing took 5-7 days.
FDE embedded across the full operations org, shadowing the daily war room for 2 weeks, visiting 3 of 9 distribution centers, and spending a week with procurement watching PO flows. Mapped 47 distinct exception types across all sub-functions.
Full operations layer spanning every sub-function from demand signal to customer delivery.
Exception detection agent monitoring SAP, Manhattan, and Blue Yonder for disruption signals and creating structured exception tickets with full pre-assembled context. Exception routing agent handling the 31 repeatable types autonomously. Escalation agent managing the 16 complex types by assembling context, recommending 2-3 resolution options with trade-off analysis, and routing to the right decision-maker.
Structured clinical notes agent running continuous forecast updates by correlating internal signals (POS data, order pipeline, returns trends) with external signals (weather, port congestion, raw material pricing, competitor activity). S&OP preparation agent auto-generating the monthly package with gap analysis, risk flags, and recommended actions.
Intelligent PO agent validating MRP-suggested quantities against real-time demand signals, supplier lead time variability, and open PO status. Supplier management agent maintaining unified scorecards from Ariba, Manhattan, and quality data. Contract compliance agent monitoring for off-contract spend and missed rebate thresholds.
Inbound visibility agent pulling real-time shipment status from carrier APIs and supplier portals, feeding into labor planning. Allocation optimization agent running continuous allocation instead of nightly batch. Returns disposition agent auto-classifying returned items and routing to restock, refurbish, or scrap.
Quality signal agent connecting quality hold data to supplier scorecards, identifying trending defect patterns, and triggering corrective action requests. Feeding quality data into procurement so sourcing decisions incorporate quality performance alongside price and lead time.
Month-by-month focus and results.
This engagement produced both a one-time working capital release and ongoing annual value. We separate them here because they compound differently: the working capital release is realized once, while the recurring savings grow as agent accuracy improves.
Recurring Annual Value
One-Time vs. Recurring: The $9.3M working capital release is a one-time benefit from reducing inventory buffers that compensated for slow disruption response. It does not recur annually but the freed capital is permanently available. Recurring value of $10.1-13.7M represents ongoing annual savings that compound as agent accuracy improves.
Ops agents fed supplier reliability scores into finance AP prioritization. Finance fed spend concentration risk back into supplier diversification strategy. Structured clinical notes data fed into sales forecasting. By month 10, three departments operated off a shared intelligence layer. These cross-department effects are not included in the value estimates above.
Revenue: €100M | 300 employees | hospital ops
Ops: consult notes, structured file, medical letter
Tech Stack: existing clinical workflow
Supply chain exception management was entirely manual. When a disruption hit, the ops team had to check SAP for inventory, Manhattan for warehouse allocation, Blue Yonder for demand forecast, then call or email 3-4 people to coordinate the response. Average disruption response time was 2-4 days.
Stockout rate was 6.8%. The ops team ran a daily war room where 12 people spent 90 minutes going through an Excel tracker, mostly status updates. Supplier performance data existed across 3 systems with no unified view. The ops team was the integration layer between these systems. When someone quit, their institutional knowledge walked out with them.
Demand planning ran a monthly S&OP cycle that was already outdated by publication. Warehouse allocation was a nightly batch process. MRP-generated POs had a 23% amendment rate. Inbound logistics was tracked in a spreadsheet by 4 people. Quality data was disconnected from supplier scorecards. Returns processing took 5-7 days.
FDE embedded across the full operations org, shadowing the daily war room for 2 weeks, visiting 3 of 9 distribution centers, and spending a week with procurement watching PO flows. Mapped 47 distinct exception types across all sub-functions.
Full operations layer spanning every sub-function from demand signal to customer delivery.
Exception detection agent monitoring SAP, Manhattan, and Blue Yonder for disruption signals and creating structured exception tickets with full pre-assembled context. Exception routing agent handling the 31 repeatable types autonomously. Escalation agent managing the 16 complex types by assembling context, recommending 2-3 resolution options with trade-off analysis, and routing to the right decision-maker.
Structured clinical notes agent running continuous forecast updates by correlating internal signals (POS data, order pipeline, returns trends) with external signals (weather, port congestion, raw material pricing, competitor activity). S&OP preparation agent auto-generating the monthly package with gap analysis, risk flags, and recommended actions.
Intelligent PO agent validating MRP-suggested quantities against real-time demand signals, supplier lead time variability, and open PO status. Supplier management agent maintaining unified scorecards from Ariba, Manhattan, and quality data. Contract compliance agent monitoring for off-contract spend and missed rebate thresholds.
Inbound visibility agent pulling real-time shipment status from carrier APIs and supplier portals, feeding into labor planning. Allocation optimization agent running continuous allocation instead of nightly batch. Returns disposition agent auto-classifying returned items and routing to restock, refurbish, or scrap.
Quality signal agent connecting quality hold data to supplier scorecards, identifying trending defect patterns, and triggering corrective action requests. Feeding quality data into procurement so sourcing decisions incorporate quality performance alongside price and lead time.
Month-by-month focus and results.
This engagement produced both a one-time working capital release and ongoing annual value. We separate them here because they compound differently: the working capital release is realized once, while the recurring savings grow as agent accuracy improves.
Recurring Annual Value
One-Time vs. Recurring: The $9.3M working capital release is a one-time benefit from reducing inventory buffers that compensated for slow disruption response. It does not recur annually but the freed capital is permanently available. Recurring value of $10.1-13.7M represents ongoing annual savings that compound as agent accuracy improves.
Ops agents fed supplier reliability scores into finance AP prioritization. Finance fed spend concentration risk back into supplier diversification strategy. Structured clinical notes data fed into sales forecasting. By month 10, three departments operated off a shared intelligence layer. These cross-department effects are not included in the value estimates above.
