Frequently Asked Questions

A typical enterprise implementation follows a structured 4-8 week timeline, depending on your operational complexity and SAP environment configuration:

• Week 1-2: Discovery & SAP data integration setup—we connect to your SAP backend and extract all work orders, operations, task lists, and historical constraints
• Week 2-4: AI model training on 15-20 historical shutdowns using Market Basket Analysis to learn precedence relationships and optimal patterns
• Week 4-6: Validation, testing, and user acceptance with your scheduling team
• Week 6-8: Production deployment, staff training, and first live optimization

For organizations seeking faster results, our accelerated pilot program can demonstrate initial optimization within 30 days on a single shutdown—replacing what would normally take your team 4 weeks of manual planning with a 5-minute AI-generated schedule.

Our AI learns your operations through a three-layer constraint learning system:

• Historical Pattern Analysis: We analyze 15-20 past shutdowns using Market Basket Analysis—the same technique retailers use to understand buying patterns. Our AI identifies which operations always occur together, which must happen sequentially, and which can run in parallel.
• Asset Management Rules: We extract constraints from your SAP task lists, maintenance plans, and asset management configurations automatically.
• Scheduler Knowledge Capture: When your planners make manual adjustments, those constraints are locked into our database. The next optimization run incorporates all scheduler-added constraints—making the system smarter with every use.

This approach achieves 89% accuracy in predicting task relationships, and the system continuously improves as schedulers validate and refine outputs.

Yes, absolutely. We strongly recommend our Proof of Value (POV) pilot program. Here's how it works:

• Deploy OptiSchedule AI on a single shutdown using your actual SAP data
• Compare our AI-optimized schedule against your current planning approach
• See real labor reduction numbers—our deployments consistently show 30-54% workforce reduction
• Typical pilot duration: 4-6 weeks

For optimal AI model training, we recommend access to 15-20 historical shutdowns. This data allows our machine learning algorithms to:

• Extract precedence and succession relationships between thousands of operations
• Build probabilistic models of task dependencies using Market Basket Analysis
• Learn which operations never occur before others, which always follow, and which can safely run in parallel
• Achieve 89% accuracy in predicting operational relationships

The AI studies patterns like: "89% of the time when crews complete vessel lowering, welding starts within 4 hours." These learned patterns become constraints that ensure realistic, executable schedules.

If you have fewer than 15 historical events, we can still proceed using industry best practices and manual constraint configuration, with the AI improving continuously as more data becomes available.

OptiSchedule AI transforms your scheduling team from manual planners into strategic optimizers:

• Before: Teams of 7-8 schedulers spend 4 weeks manually creating schedules, juggling 500-600 constraints in their heads
• After: AI generates the optimized baseline in 5 minutes with zero constraint violations
• New Role: Schedulers validate AI output, handle exceptions, and focus on strategic improvements

Organizations typically see 60-80% reduction in scheduling staff requirements. In one deployment, a team of 250 schedulers was reduced to 50—because you don't need 4 weeks of manual planning when the AI delivers in 5 minutes.

The remaining schedulers become more valuable, focusing on exception handling, continuous improvement, and strategic planning rather than repetitive manual work.

Industrial shutdown scheduling involves an astronomical number of possible combinations—10²⁸⁸ possibilities. To put this in perspective: brute-forcing this with a quantum computer would take 10²⁵ years, while the universe is only 1.3 × 10¹⁰ years old.

OptiSchedule AI solves this using a multi-technology approach:

• Constraint Programming: Mathematical modeling that eliminates impossible combinations before computation
• Tree Pruning Algorithms: When the AI explores solution branches, it identifies paths where optimization won't occur and cuts them—eliminating trillions of combinations instantly
• Massively Parallel Processing: Server clusters try different combinations simultaneously across distributed infrastructure
• Machine Learning: Historical patterns guide the search toward optimal solutions faster

The result: 2,000+ operations scheduled in 5 minutes with zero constraint violations—work that takes human planners 4 weeks.

Mid-shutdown disruptions are inevitable—equipment failures, break-in work, weather delays. Traditional replanning takes 2-3 days. OptiSchedule AI does it in 20 seconds.

Here's how it works:

• Break-in Work: You're halfway through a shutdown and 5 additional work orders surface. Input them into the system.
• Instant Re-optimization: Within 20 seconds, the AI re-evaluates the entire schedule considering all existing constraints
• Best Placement: The system shows exactly where to insert the new work orders with minimum labor impact
• Real-Time Labor Impact: See immediately how much additional labor is required and timeline implications
• Accept or Adjust: Approve the recommendation or drag-and-drop to modify, with constraint violation warnings

This eliminates the chaos of emergency replanning and the expensive last-minute labor at triple-rate pay that comes when you can't rebalance quickly.

OptiSchedule AI handles the real-world complexity that makes manual scheduling so difficult:

• Physical Constraints: Work on upper vessels (100-200 meters high) blocks work below them. The AI ensures proper sequencing automatically.
• Crew Dependencies: Until one crew lowers a vessel, welding crews can't begin. Until welding completes, assembly crews can't start. These chains are enforced.
• Resource Constraints: 101 types of mechanical maintainers, each with different certifications. The system tracks exactly which type you need for each half-hour slot.
• Accommodation Limits: Site maximum occupancy (e.g., 1,500 people). The AI flattens labor curves to stay within capacity.
• Shift Patterns: Day shift 6AM-6PM, night shift 6PM-6AM. Work never stops, and the AI optimizes across both.
• Ramp-Up/Ramp-Down: Inspection jobs during plant shutdown and startup phases are automatically placed in correct pockets.

All constraints are auto-enforced. If you try to override one manually, the system warns you with red indicators showing exactly which constraints you're violating.

Yes—this is exactly what we're built for. Our production deployment spans:

• 27-28 sites across Western Australia
• Multiple plants per site with different maintenance schedules
• Shutdown maintenance (7-14 days, 4 times per year per site)
• Weekly primary maintenance (PMO1)—essentially 52 mini-shutdowns per year per site

The system handles:

• Site Selection: Choose your site, choose your shutdown, select feed-off time and tool start time
• Cross-Site Resources: Optimize labor sharing between nearby sites
• Charter Flight Logistics: When accommodation is limited, optimize crew rotation schedules
• Centralized Visibility: Enterprise-wide view with drill-down to individual site schedules

Discover our Mining Scheduling solution designed for multi-site operations.

Yes—this is a core design principle. The system improves through continuous learning:

• Constraint Locking: When a scheduler says "This operation must be in day shift" or "This predecessor relationship is wrong"—those constraints are locked in a database
• Re-optimization Learning: The next time you run the optimizer, it considers all previously locked constraints automatically
• Pattern Refinement: Each completed shutdown adds to training data, improving Market Basket Analysis accuracy
• Feedback Loop: If schedulers consistently make the same type of adjustment, the AI learns to anticipate it

After 5, 10, 15 corrections, the system becomes intelligent enough that its output already incorporates what schedulers would have done manually. Theoretically, a stage comes when manual intervention is minimal.

Our development team continuously updates the system based on scheduler feedback—it's been getting better every month for the past 6-8 months of production use.

OptiSchedule AI is built specifically for SAP environments. Our integration works like this:

• Data Extraction: We connect to your SAP backend and extract all work orders, operations, task lists, maintenance plans, and historical constraints from relevant SAP tables (AFKO, AFVC, CRHD, etc.)
• Historical Learning: The system pulls 15-20 past shutdowns to train the AI on your specific operational patterns
• Optimization: All constraints, precedence relationships, and resource requirements are processed through our AI engine
• Writeback: Optimized schedules are written directly back to SAP with full audit trail

Key advantage: SAP's native frontend is weak—there's no interactive Gantt chart for 8,000+ work orders. Companies pay competitors like Prometheus $8 million per year just for basic drag-and-drop UI on top of SAP. OptiSchedule AI provides that UI plus AI optimization that Prometheus can't offer.

Learn more about our SAP Integration solution.

SAP's native scheduling interface is fundamentally limited. When you have 4,000-8,000 work orders:

• SAP Limitation: No interactive Gantt chart. Schedulers manually drag and drop operations one by one, trying to remember 500-600 constraints in their heads.
• Our Solution: A fully automated Gantt chart that displays all work orders and operations—automatically planned with zero manual intervention

Our interactive GUI provides:

• Drag-and-Drop Editing: Move any work order and instantly see labor impact below the chart
• Real-Time Constraint Warnings: Red indicators appear immediately when you violate any constraint
• Undo Capability: Made a mistake? Undo any changes instantly
• Constraint Override Options: Choose to follow or ignore specific constraints with full visibility
• SAP Writeback: Finalized schedules sync directly back to SAP

This replaces tools that charge $1 million+ for basic "start from zero and drag" functionality with intelligent optimization included.

Yes. While our core system is optimized for SAP, we support multiple enterprise platforms:

• Oracle Primavera P6: Project scheduling synchronization for construction and turnaround projects
• IBM Maximo: Asset management and work order integration
• Microsoft Project: Desktop and Project Online/Server
• Custom ERP Systems: API-based integration through REST endpoints

For companies like Saudi Aramco or other oil & gas majors using SAP—the integration is straightforward because they also use SAP. The AI goes into their history, analyzes previous maintenance events, and optimizes based on their specific patterns.

Export formats include Excel, PDF Gantt charts, and Primavera XML for seamless data sharing.

The system handles the full spectrum of industrial maintenance:

• Shutdown Maintenance: Major 7-14 day turnarounds, typically 4 times per year. This is where the biggest savings occur.
• Primary Maintenance (PMO1): Weekly maintenance that happens continuously—essentially 52 mini-shutdowns per year. Just select time window and press optimize.
• Preventive Maintenance: Filter changes every 2 weeks, oil changes at intervals—maintenance plans deployed in SAP. The AI automatically schedules this work.
• Break-In Work: Unplanned work that surfaces during operations. Add it mid-shutdown and get re-optimized schedule in 20 seconds.

The weekly primary maintenance opportunity is actually bigger than shutdowns—it happens every week versus 4 times per year. This is where reliability metrics are maintained and equipment runs "practically forever."

Learn more about our Shutdown Scheduling solution and AI Scheduling capabilities.

Based on verified production deployments across 27 shutdowns, here are the actual results:

• Workforce Reduction: 700 workers reduced to 320—a 54% reduction in a single shutdown
• Peak Labor Flattening: Peak headcount dropped from 100+ people to never exceeding 65
• Average Labor Savings: 30-50% reduction across most deployments
• Planning Time: 4 weeks of manual scheduling reduced to 5 minutes
• Scheduler Requirements: 250 schedulers reduced to 50

Explore our Workforce Optimization solution for more details.

The financial impact scales with operational size. Here are conservative estimates from production deployments:

• $1.8M+ saved per shutdown through workforce optimization
• $50 million annual savings projected for enterprise-wide deployment
• $27 million per week in operational inefficiency eliminated
• Scheduler cost savings: 200+ schedulers no longer needed for 4-week planning cycles
• Charter flight elimination: Hundreds of thousands saved when accommodation constraints are solved through labor flattening

Additional savings come from:

• Eliminated triple-rate emergency labor during replanning crises
• Reduced accommodation and logistics costs from flatter labor curves
• Shortened shutdowns = more production days

Labor spikes are the hidden cost killer in manual scheduling. Here's what happens:

• The Problem: Manual schedules create peaks where you need 60 mechanical maintainers for 2 shifts, but only 20 for the rest. You hire 60, pay for 7 days, but they sit idle 5 of those days.
• The Solution: OptiSchedule AI sequences work orders so labor consumption becomes completely flat—no spikes.

Real results:

• Day/night shift mechanical maintenance: never exceeds 16 people (was spiking to 100+)
• Total headcount stays around 118 people instead of 1,186
• Buy 30 workers and work them for 7 days instead of buying 60 and paying them to sit idle

The AI shows exactly which trade type you need for each half-hour slot—mechanical maintainers (101 different types), electricians, loop technicians, trade assistants, scaffolders, crane operators—all with half-hour resolution.

This is a critical pain point that manual scheduling can't solve:

• The Conflict: Third-party labor companies' job is to sell labor. They want to provide 10 people. The supervisor's KPI is to turn the plant back on in 7 days—so they accept whatever labor is quoted to avoid risk.
• The Problem: Supervisors have no tool to argue with contractors. "I don't need 10 people, I need 2" has no backing data.

OptiSchedule AI provides the data:

• Half-Hour Resolution: Exactly how many people of each trade type you need for every 30-minute slot
• Shift-Level Detail: "Night shift—you need only 3 electricians for this work center, 1 for that one"
• Trade Breakdown: Mechanical maintainer: 12 for work center A, 7 for B, 1 for C—with specific certifications

Now supervisors can say: "The system shows I need 2 people, not 10. Here's the data." This shifts the negotiation power entirely.

OptiSchedule AI is built on elastic cloud infrastructure designed to handle computationally intensive optimization:

• Server Clusters: When you press "Optimize," the system spins up a bank of multiple servers that try different combinations in parallel
• Auto-Scaling: Infrastructure scales up during optimization runs and scales down when complete—you only pay for what you use
• Distributed Processing: Massively parallel computation across cloud server clusters enables exploring billions of combinations in 5 minutes
• Regional Deployment: Data centers positioned near operational sites for low-latency access

The cloud architecture is specifically designed for the compute-heavy nature of constraint programming and tree pruning algorithms—processing that SAP itself acknowledged they "can't do because it's very compute-hungry."

We work with major mining and energy companies who require the highest security standards:

• Data Encryption: AES-256 encryption at rest and TLS 1.3 in transit
• Tenant Isolation: Complete data isolation—your operational data is never commingled with other customers
• Access Controls: Role-based access with multi-factor authentication
• Audit Logging: Complete audit trail of all system access and schedule changes
• SAP Security: Integration uses secure RFC/BAPI connections with your existing SAP authentication

Our security framework meets the requirements of Fortune 500 mining operations where operational data is commercially sensitive.

Yes. For organizations with strict data residency or air-gap requirements, we offer flexible deployment options:

• Public Cloud (Standard): Fully managed SaaS deployment on AWS/Azure with elastic scaling
• Private Cloud: Dedicated infrastructure in your preferred cloud provider's region
• On-Premises: Full deployment within your data center—all computation happens behind your firewall
• Hybrid: SAP connectivity on-premises, optimization in cloud, results synced back

Remote mining and oil & gas sites often have connectivity constraints. Our architecture supports offline optimization with sync-when-connected capabilities.

Human oversight is built into the core workflow. Here's how it works:

• Rough Cut Delivery: The AI generates a complete schedule with 2,000+ operations in 5 minutes—this is your "rough cut" baseline
• Critical Path Alerts: If any work orders don't fit within the prescribed timeline (even with all constraints followed), the system tells you: "Fix these yourself"
• Drag-and-Drop Editing: Move any work order in the interactive Gantt chart
• Real-Time Labor Impact: As you drag, see exactly how labor requirements change below the chart
• Constraint Warnings: Red indicators immediately show which constraints you're violating
• Full Undo Capability: Made a mistake? Undo any and all changes

The AI does 75-80% of the work. Schedulers validate, adjust the remaining edge cases, and lock in constraints that improve future runs.

This is the critical question every supervisor asks: "The system says 320 workers, but if even one day goes over, my KPI is ruined."

We address this through a confidence-building approach:

• Start Conservative: If the system says 320 and you have accommodation for 400, start with 400. The numbers still look dramatically better than 700.
• Track Accuracy: After each shutdown, compare actual vs. planned. As supervisors see "this works correctly every time," confidence grows.
• Buffer Capability: The 20-second replanning means if something goes wrong mid-shutdown, you can instantly rebalance rather than panic-hiring at triple rates.
• Continuous Improvement: The system learns from scheduler feedback. Over 6-8 months of use, it gets better and better.

One manager's response: "Instead of 700, this is saying 320, which is correct. But let me try 400 just in case." That's still a massive improvement, and confidence builds from there.

OptiSchedule AI is continuously improved based on real-world scheduler feedback:

• Feedback Loop: When schedulers say "add this feature" or "this constraint doesn't work"—we update the system. It's been getting better every month for 6-8 months.
• Development Team Access: Direct access to the development team for feature requests and customizations
• Zero-Downtime Updates: System improvements deployed without service interruption
• Training Resources: Comprehensive documentation, training sessions, and ongoing support

Enterprise support includes:

• 24/7 coverage for critical shutdown periods
• Dedicated success manager for your account
• 1-hour response SLA for critical issues
• 99.9% uptime SLA with redundant infrastructure