Executive Summary
A Fortune 500 industrial gases manufacturer headquartered in Pennsylvania with $12.3B in annual revenue and 22,000+ employees eliminated stockout events across 340+ hospital and manufacturing customer accounts by deploying AI-driven telemetry replenishment automation.
The company reduced emergency fill costs by 100%, optimized inventory carrying costs by 18%, and achieved full ROI within seven months while improving customer retention scores by 23 percentage points.
Retrun on Investment
Elimination of Emergency Fills
Reduction in Inventory Carrying Cost
SITUATION
The client operates as one of the top three industrial gas suppliers in North America, serving critical sectors including healthcare (hospital oxygen systems), food processing (nitrogen freezing), and advanced manufacturing (specialty gas blends). With distribution operations spanning 180+ facilities across the continental United States, the company manages bulk gas supply contracts for over 2,400 commercial accounts. Of these 2,400 commercial accounts, 340+ mission-critical accounts (primarily hospitals and food processors requiring guaranteed supply continuity) were selected as the initial automation cohort for the IQ Platform deployment.
Their traditional business model involved on-site bulk storage tanks at customer facilities, ranging from 500-gallon liquid oxygen dewars at regional hospitals to 13,000-gallon liquid nitrogen tanks at food processing plants. Prior to automation, the company had begun deploying IoT telemetry sensors on approximately 40% of customer tanks (roughly 960 installations), providing real-time pressure and consumption data transmitted via cellular networks to their operations center in Trexlertown, PA.
Despite having telemetry infrastructure in place, the company still relied heavily on manual processes for replenishment planning. A team of 14 supply chain coordinators monitored dashboards showing tank levels across hundreds of accounts, using static min/max thresholds (typically set at 20% minimum, 90% maximum) established during contract negotiation. The existing tech stack included SAP ERP for order management, Salesforce for customer relationship management, and a legacy logistics system for fleet dispatch, but these systems operated in silos with minimal integration.
TRIGGER
In Q2 2023, the company experienced a cascade of service failures that threatened major contract renewals. A severe winter storm in the Midwest disrupted normal delivery schedules, causing simultaneous tank depletion at three hospital systems within a 72-hour window. One facility, a Level I trauma center in Ohio, came within four hours of exhausting its oxygen supply before an emergency fill could be executed.
The incident triggered regulatory scrutiny from state health departments and prompted the hospital network (representing $18M in annual contract value) to issue a formal breach notice. Internal post-mortem analysis revealed that static thresholds had failed to account for increased oxygen consumption during flu season, compounded by weather-related delivery delays. The company’s VP of Supply Chain Operations faced pressure from executive leadership to prevent future stockouts while simultaneously reducing the spiraling costs of emergency fills, which had reached $840,000 in the previous fiscal year.
Competitive pressure intensified as two rival industrial gas suppliers began marketing “predictive replenishment” capabilities in their RFP responses. The trigger was clear: manual monitoring could no longer support the reliability standards demanded by mission-critical customers, and the cost structure of reactive replenishment was eroding margins.
BARRIER
The company confronted multiple operational and technical obstacles that prevented optimization of their telemetry infrastructure:
Data Volume Overload: 14 supply chain coordinators manually reviewed telemetry dashboards for 960+ IoT-enabled tanks, with each coordinator responsible for monitoring 68+ accounts, an impossible workload resulting in delayed identification of at-risk tanks.
Static Threshold Failures: Min/max replenishment triggers (20%/90%) ignored demand variability; hospital oxygen consumption fluctuated 35-60% seasonally (winter respiratory illness spikes), while food processing nitrogen demand varied with production schedules.
Lead Time Blindness: Existing thresholds didn’t factor in delivery lead times, which ranged from 24 hours (urban hospitals) to 96+ hours (rural manufacturing sites during winter months); coordinators couldn’t predict when tanks would reach critical levels relative to delivery windows.
System Integration Gaps: IoT telemetry platform (GE Predix), SAP ERP, and fleet management system operated as disconnected silos; coordinators manually transferred data between systems, creating 4-6 hour delays between alert identification and PO generation.
Emergency Fill Cost Drain: Unplanned emergency replenishments cost $2,200-$3,800 per event (off-route dispatch, overtime labor, expedited logistics); the company executed 280+ emergency fills annually, totaling $840,000 in avoidable costs.
Inventory Inefficiency: Conservative safety stock policies (overfilling to compensate for uncertainty) increased carrying costs and reduced delivery efficiency; trucks often traveled with partial loads or made redundant trips to the same geographic clusters.
During pilot testing in Week 8, the team discovered that legacy GE Predix API rate limits (500 calls/hour) couldn’t support real-time data pulls for 960 tanks.
Solution
The company selected the IQ Platform, implemented by UTOFA, as the orchestration layer for telemetry-driven replenishment automation, chosen specifically for its event-driven workflow architecture and ability to integrate disparate enterprise systems without requiring full replacement of legacy IT infrastructure. Competing solutions from traditional RPA vendors required extensive custom coding and lacked the real-time event processing capabilities needed for time-sensitive replenishment decisions.
Why IQ Platform:
The client chose IQ Platform over incumbent RPA vendors (UiPath, Blue Prism) due to three critical factors: 1. event-driven architecture enabled sub-second response to telemetry threshold breaches versus 15-30 minute polling cycles in traditional RPA; 2. SLM-based reasoning engine could incorporate unstructured inputs (weather alerts, customer notes) without requiring extensive retraining; 3. 16-week deployment timeline versus 9-12 month estimates from SAP IBP full implementation.
IoT Data Ingestion: IQ Platform’s Stateful Event Handling System integrated with existing GE Predix telemetry infrastructure via REST API, ingesting real-time tank level data (pressure, flow rate, temperature) at 15-minute intervals. To overcome GE Predix API rate limits (500 calls/hour), IQ Platform implemented a delta-polling strategy, querying only tanks with status changes rather than full sweeps, reducing API calls by approximately 73% while maintaining effective real-time coverage across all 960 tanks.
Demand Forecasting Engine: Prophet-based ML model trained on 18 months of historical consumption data, incorporating variables including seasonal patterns, day-of-week effects, customer industry segment, and regional weather forecasts (integrated via NOAA API).
Dynamic Threshold Optimization: IQ Platform’s Cognitive Workflow Management calculated customer-specific replenishment triggers based on forecasted consumption rate, delivery lead time (pulled from fleet management system), and safety buffer (industry-specific: 15% for hospitals, 10% for manufacturing).
Automated Order Generation: When tank level projections indicated minimum threshold breach within lead time window, IQ Platform’s Intelligent Orchestrator automatically generated purchase orders in SAP ERP, including delivery priority scoring (emergency/standard/optimized).
Route Optimization Integration: IQ Platform coordinated with Descartes fleet management system to optimize delivery routes; grouped geographically proximate customers into multi-stop routes when timing allowed.
The Results
The IQ Platform implementation delivered transformational results within the first 18 months by replacing fragmented legacy systems with intelligent orchestration:
Financial Impact: $4.7M in total annualized savings: $840K emergency fill elimination + $2.1M inventory optimization + $980K labor reallocation + $780K route efficiency gains.
Operational Excellence: 100% stockout elimination with zero emergency fills in 18 months post-deployment across all automated accounts. Cycle time reduced from 4-6 days (manual monitoring to delivery) to 24-36 hours (automated trigger to delivery).
ROI Achievement: 267% ROI with 7-month payback on $125K platform investment plus $18K monthly managed services.
Inventory Optimization: 18% reduction in inventory carrying costs through dynamic threshold optimization; safety stock reduced from 25% average to 12% average tank capacity.
Customer Satisfaction: 23-point NPS improvement driven by proactive delivery notifications and zero service interruptions.
"The IQ Platform transformed our supply chain from reactive firefighting to predictive orchestration. We've eliminated emergency fills entirely while reducing inventory costs by nearly 20%, outcomes our legacy systems couldn't deliver despite years of investment."
"What separated this implementation from previous automation projects was speed and integration depth. We went from pilot to full production in four months, and the platform integrated with our existing SAP and Predix infrastructure without requiring us to rip and replace systems."
Summary
- Achieved 7-month ROI with 267% return delivering $4.7M in annual savings through emergency fill elimination ($840K), inventory optimization ($2.1M), labor reallocation ($980K), and route efficiency gains ($780K)
- Eliminated 100% of emergency fills maintaining zero stockout events across 340+ mission-critical accounts over 18 months through predictive replenishment orchestration
- Reduced replenishment cycle time by 75% from 4-6 days (manual monitoring to delivery) to 24-36 hours (automated trigger to delivery) through real-time telemetry integration
- Optimized inventory carrying costs by 18% reducing safety stock from 25% to 12% average tank capacity through dynamic threshold optimization and demand forecasting
- Improved customer retention scores by 23 percentage points through proactive delivery notifications and zero service interruptions across hospital and manufacturing accounts
- Overcame API rate limitations reducing GE Predix API calls by 73% through delta-polling strategy while maintaining real-time coverage across 960 IoT-enabled tanks
Key Metrics
| KPI | Before Automation | After Automation | Improvement |
|---|---|---|---|
| Emergency Fills (Annual) | 280 events | 0 events | 100% reduction |
| Average Replenishment Lead Time | 4–6 days | 24–36 hours | 75% reduction |
| Inventory Carrying Cost (% of Revenue) | 3.2 % | 2.6 % | 18% reduction |
| Supply Chain Labor (FTEs) | 14 coordinators | 6 coordinators | 57% reallocation |
| Customer NPS Score | 52 | 75 | +23 points |
| Route Efficiency (Deliveries/Truck/Day) | 3.4 | 4.8 | 41% increase |
