Inside Jabil's Integration-First Playbook for Scaling AI Across 100+ Manufacturing Sites

If you've ever inherited a codebase held together by cron jobs, spreadsheets, and three generations of "temporary" workarounds, you'll recognize Jabil's problem immediately. It's just running at industrial scale.

Jabil is a global manufacturing company with 60 years in business, 140,000-plus employees, and more than 100 sites spread across 30-plus countries. It builds for over 400 of the world's top brands. That kind of footprint doesn't happen overnight, and neither does the technical debt that comes with it. Over 25 years, individual sites accumulated their own tools, their own workarounds, their own spreadsheet-based processes to plug gaps that never got properly engineered. The result: a patchwork of disconnected systems that made it genuinely hard to spot problems early or coordinate a response across regions.

Harish Manohar, SAP IT director at Jabil, describes the company's response as a "simplify-first, then-innovate mindset." It's a deceptively simple framing, but it's the kind of discipline a lot of engineering orgs talk about and few actually enforce. "Any innovation without simplification is going to add more complexity," Manohar says. Before Jabil would touch AI-driven planning or predictive supply chain tools, it needed to fix the plumbing.

That plumbing problem is one every practitioner will recognize. You can't run good models or good automation on bad, siloed, inconsistent data. Manohar puts it bluntly: "the backbone of any contemporary or modern organization is data." Optimization, automation, AI, none of it works if data can't flow seamlessly between systems first.

The Architecture Behind the Simplification

Jabil's technical approach centers on SAP's Business Technology Platform (BTP) and Integration Suite, which the company is positioning as the central nervous system for its integrations. It's not there yet, by Manohar's own admission, but the direction is clear: consolidate around one integration platform rather than maintaining a sprawl of point solutions, and reserve smaller tools for cases where they genuinely add value.

A few implementation details stand out for anyone doing similar infrastructure work:

• API-driven, event-based integration is the standing architectural policy. Jabil explicitly avoids batch-style data shuffling between systems. If you've ever debugged a nightly ETL job that silently failed and corrupted three downstream systems before anyone noticed, you understand why event-driven wins here.
• Reuse over one-off tooling. Rather than buying a new point solution every time a site has a specific need, Jabil tries to standardize on what already exists elsewhere in the network, adjusting business process rather than adding another vendor to the stack.
• Clean-core alignment with SAP RISE. This is arguably the most consequential decision. Jabil's SAP environment has been heavily customized for 25 years to satisfy customer-specific demands. Heavy customization is a classic trap: it solves today's problem while making every future upgrade slower and riskier. Jabil is now moving to RISE (SAP's managed cloud transition path) specifically to enable a "clean core," meaning fewer bespoke modifications and stronger governance over what gets customized going forward.
• Signavio for process standardization. Jabil uses SAP Signavio, a business process management tool, to help standardize how its 100-plus plants operate despite differing regulatory environments and process maturity levels.

The clean-core piece deserves a beat of its own. Anyone who's worked in enterprise software knows customization debt compounds. Every one-off modification you ship today is a future migration headache, a merge conflict with the vendor's roadmap, a reason your next major upgrade takes six months instead of six weeks. Jabil is essentially doing a large-scale refactor of its SAP estate, tightening governance so new customizations require real justification rather than defaulting to "just build it."

None of this rollout is trivial given Jabil's operating reality. Its sites vary wildly in process maturity, legacy tooling, and local workflow quirks. Some of its business units are regulated, which adds qualification and compliance overhead (CSV processes, in Manohar's phrasing) on top of the usual integration challenges. Standardizing across that landscape means changing process and governance without breaking operations that are actively running production lines. That's the software equivalent of refactoring a live system with zero downtime tolerance.

Manohar frames the payoff in concrete operational terms: standardization means consistent workflows, data flows, and governance across sites, and it materially speeds up future tech rollouts. Historically, deploying something like BTP or Signavio to a new site meant navigating heavy site-specific customization first. Strip that out, and rollout velocity goes up. Jabil's initial target is scaling shared, consistent processes across 40-plus plants, a meaningful chunk of its total footprint.

There's also a human dimension worth flagging, since it's easy to lose in a conversation about pipelines and APIs. Integrated workflows mean employees get shared visibility into data instead of chasing it across disconnected systems. Less manual reconciliation, more time spent acting on actual insight. For a manufacturing operation, that translates into faster response to supply chain disruptions and, ideally, lower operational risk when something breaks on the line.

Key Takeaways

The AI ambitions here are still on the horizon rather than in production at scale. With trusted data and integrated systems as the foundation, Jabil is exploring predictive supply chain insights, intelligent exception handling, and AI-driven planning and forecasting. Notably, none of that comes first. It comes after the integration work, not instead of it.

For engineers watching enterprise AI hype cycles from the outside, Jabil's approach is a useful counterpoint to the "bolt an LLM onto it" instinct. Manohar's closing line is the whole thesis in miniature: "Simplicity at scale is a very competitive advantage." Technology investment only matters if it ties back to measurable business value and operational resilience, and that starts with unglamorous integration work long before any model gets near production data.