Executive Summary
A Fortune 500 global commercial vehicle manufacturer set up an AI-first Global Capability Center (GCC) to modernize shop-floor and enterprise workflows. Its plants faced disconnected IT and OT data, manual component checks, error-prone capture of long serial numbers, costly dedicated scanners and a hard rule that production could not stop. The GCC tackled these in order: it connected shop-floor data to MES, EAM and SAP, put AI agents in charge of routine component validation, and moved scanning to standard Android devices. This article walks through the challenge, how the center was architected, and the lessons other manufacturers can take from it.
What was holding back AI on this manufacturer's shop floor?
The plants had plenty of data, but it was disconnected, captured by hand and checked manually, so AI had nothing reliable to work with. Five problems stood in the way.
- IT and OT data lived in separate worlds. Shop-floor logs and component verification records sat apart from the manufacturing execution system (MES), enterprise asset management (EAM) and SAP. Without IT and OT convergence, no one could see in real time whether the right part had gone into the right vehicle.
- Component checks were manual. Teams validated component age, configuration, reuse and assembly by hand. That made the checks slow, inconsistent between shifts and prone to error, with mistakes often found only after the fact.
- Serial numbers were hard to capture accurately. Operators had to record 17-digit serial numbers in fast-paced conditions. Typing them by hand led to inaccurate or incomplete records, which weakened serial number tracking and component traceability across the build.
- Dedicated scanners were costly and limiting. Specialized scanning hardware added infrastructure cost and tied operators to fixed stations, which didn’t suit a moving assembly line.
- Production could not stop. Any change had to happen while the plants kept running. A big-bang replacement of shop-floor systems was never an option, so the center had to modernize in place.
How was the AI-first GCC architected to fix it?
The center built four layers that work together: a connected data foundation, AI agents for validation, mobile edge capture and specialist engineering pods. Each one solved a specific problem from the shop floor.
- A data layer that connects the shop floor to the enterprise. Shop-floor data was normalized into a common format and linked securely, in real time, to MES and SAP. Production events and enterprise records could now describe the same component in the same way.
- AI agents that validate components automatically. Agents now check component configuration and assembly against the rules, flag incorrect assemblies as they happen, and raise work orders in EAM and CMMS when someone needs to step in. People handle the exceptions instead of every routine check.
- Offline scanning on standard Android devices. An edge SDK reads serial numbers and barcodes with OCR on ordinary Android phones, even without a network connection. Operators scan 17-digit serials instead of typing them, which improves shop floor data collection without dedicated hardware.
- Specialized engineering pods. Cross-functional pods covered edge vision, data modernization, agentic AI and Android engineering. Each pod owned its part of the solution while working to shared standards, so the pieces fit together.
- Modernization without downtime. The new layers were added around the existing MES, EAM and SAP systems rather than replacing them, so production kept running while the shop floor became digital.
What did the AI-first GCC deliver, and what can other manufacturers learn?
The center cut production disruptions, put more shop-floor data into digital systems and automated component validation. The way it got there offers practical lessons for any manufacturer planning a GCC.
- Fewer unplanned production disruptions. With incorrect assemblies flagged as they happen and work orders raised automatically, problems are caught before they stop the line. [PROOF: % reduction in unplanned production disruptions]
- More shop-floor data captured digitally. Mobile OCR scanning replaced manual entry of serial numbers, so more of what happens on the line now lands in MES and SAP accurately. [PROOF: % increase in digital shop-floor data capture]
- More component validation done automatically. AI agents now handle routine configuration and assembly checks that used to be manual, freeing quality teams to focus on the exceptions. [PROOF: % increase in automated component validation]
- Lesson: fix data capture before scaling AI. The AI agents only work because serial numbers and shop-floor events are captured accurately and linked to enterprise systems. Manufacturers that start with AI models before fixing data capture can end up with agents reasoning over incomplete records.
- Lesson: modernize around production, not through it. By connecting to existing MES, EAM and SAP rather than replacing them, and by using standard Android devices instead of new hardware, the center delivered change without halting plants or adding major infrastructure cost.
Conclusion
The most telling part of this story is where it started. Before any agent made a decision, the center made sure a serial number could be captured correctly on a moving line and matched to the right record in SAP. That groundwork is what lets AI take over routine checks with confidence, and it’s the step many AI programs in manufacturing try to skip. OptiSol architected this GCC with the manufacturer, using iBEAM to connect shop-floor data with MES and SAP, elsai to build the validation agents, and the Scanflow SDK for offline scanning on Android, delivered by engineering pods working as part of the client’s team. For manufacturers weighing their own AI-first GCC, the plant floor may be the right place to begin.
FAQs:
What is an AI-first GCC in manufacturing?
An AI-first Global Capability Center in manufacturing is an offshore engineering center set up from the start to build data, AI and automation for plants and enterprise systems, rather than to provide back-office support. In this case, the center connected shop-floor data to MES, EAM and SAP and used AI agents to validate components.
What is IT and OT convergence?
IT and OT convergence means connecting operational technology on the shop floor, such as machines, scanners and production logs, with information technology systems like MES, ERP and EAM. It gives plants one real-time view of production, which AI needs to work reliably.
How can manufacturers improve shop floor data collection?
They can replace manual entry with automated capture, such as OCR and barcode scanning on mobile devices, and connect that data directly to MES and ERP. In this case, offline scanning on standard Android devices replaced manual entry of 17-digit serial numbers.
How do AI agents help with component validation?
AI agents check component configuration and assembly against defined rules, flag incorrect assemblies as they happen, and raise work orders in EAM or CMMS when a person needs to act. This turns slow manual checks into continuous, consistent validation.
Can a manufacturer modernize shop-floor systems without stopping production?
Yes. By adding a data layer, AI agents and mobile capture around existing MES, EAM and SAP systems, rather than replacing them, a manufacturer can modernize while the line keeps running, as this commercial vehicle maker did.
