Somewhere between a manufacturer’s warehouse and a pharmacy counter in rural Southeast Asia, a chain of digital handoffs decides whether medicine arrives on time. One of Asia’s largest healthcare distribution companies orchestrates that chain across more than a dozen markets, serving hundreds of thousands of medical facilities and counting many of the world’s leading pharmaceutical companies as clients. Its cold chain and logistics backbone has moved vaccines and essential medicines at national scale across the region, including through the pandemic.
But the technology holding that operation together had a fragility problem. And the engineering team that solved it built something worth studying.
Fragile Screen-Scraping Bots at Breaking Point
The company runs on a sprawling SAP ERP environment. To automate 180 business processes spanning order-to-cash, finance, and eCommerce across its markets, the team had built roughly 180 bots on SAP iRPA. Most relied on screen-scraping, replicating human keystrokes through SAP’s user interface to extract data from PDFs and Excel files, create sales orders, process deliveries, and clear receivables. A handful of newer automations already used APIs, but the bulk of the portfolio stayed tethered to the front-end.
Functional, yes. Fragile, critically so.
A cosmetic change to any SAP screen layout could set off a cascade of failures across multiple markets. The team managing these business-critical automations found themselves troubleshooting after office hours and through weekends. Years of layered legacy systems, unwanted siloes, and in-house applications had accumulated on top of processes that were not standardised across countries.
Three compounding forces made the status quo untenable. The iRPA bots offered limited diagnostic granularity, turning every failure into an investigation. The team’s automation ambitions had outpaced what screen-scraping could deliver, and the backlog of requests had ballooned over 12 months. Most pressingly, a planned migration from SAP ECC to S/4HANA meant every screen-scraping bot in the portfolio would inevitably break.
The choice became binary: upgrade to a newer version of iRPA, or reimagine the entire architecture.
“Every SAP interface update was a risk event. Our bots would break, and we’d spend entire weekends diagnosing failures across multiple markets. We needed an architecture that could absorb change, not shatter under it.”
— Engineering Manager leading the migration
A Translation Layer Built for Orchestration, Not Imitation
The engineering team chose the bolder path. Rather than rebuilding bots that mimicked human behaviour on SAP’s front-end, they partnered with Workato to construct a cloud-first orchestration layer connecting directly to SAP through APIs and backend integration points.
The conceptual shift matters. Screen-scraping is reactive and surface-level: it shatters when interfaces change. Orchestration is structural. It operates at the data and process layer, indifferent to cosmetic shifts in the UI.
The team established a Centre of Excellence and designed a “translation layer”: a standardisation and orchestration tier that absorbs the heterogeneity of its many markets and converts localised workflows into governed, repeatable automation patterns, without vendor lock-in.
The migration is well underway, with the bulk of the portfolio already re-platformed onto Workato. Where staff once transcribed data from PDFs and spreadsheets by hand, Workato now orchestrates with AI to extract information and create sales orders and deliveries directly in SAP. The Accounts Receivable (AR) clearing workflow is one example: payments are ingested from emails and MT942 files, payment information is extracted using AI services, validated and matched against customer invoices in SAP, cleared via SAP’s Business API, and anomalies are routed to the business team through Microsoft Teams.
AR clearing is one of many automations already migrated. Others include purchase order processing, high-volume sales order processing handling several thousands of emails per day, return order processing, and promotions loading, each running in a different market.
Speed, Scale, and Pharmacies That Wait Less
The functional gains are immediate. Workato’s built-in logging, error handling, and retry capabilities surface problems in minutes that once consumed entire weekends, and the migration removes the risk that S/4HANA would break the portfolio.
The business gains compound. Cycle times have dropped, infrastructure dependencies and their costs have shrunk, and the 12-month automation backlog is clearing without platform maintenance acting as a bottleneck. Reusable components enforce governance at the platform level rather than case by case.
The human gain is the one the team notices most. Engineers moved from firefighting on weekends to building new automations with confidence.
- From weekends to minutes: Issues that once consumed entire weekends now surface in minutes through built-in observability, such as detailed logging and execution monitoring.
- Migration risk neutralised: The move to a cloud-first orchestration layer removed the threat that the S/4HANA upgrade would break screen-scraping bots.
- Backlog clearing: A 12-month automation backlog is being worked down without platform maintenance as a bottleneck.
- Governance by design: A reusable component library enforces standards across every market.
“The difference isn’t just technical. Our team went from firefighting on weekends to building new automations with confidence. That shift in energy is what scales.”
– Engineering Manager leading the migration
The downstream effect is concrete. Pharmacies across the region receive medicines faster, and for underserved communities in remote areas, that acceleration means medicine reaching patients when they need it.
Scaling the Centre of Excellence
With a governed architecture in place and the S/4HANA migration risk neutralised, the Centre of Excellence is positioned to extend orchestration into new process domains. The reusable component library the team has built creates a foundation for compounding returns on every new workflow. Their ambition now points beyond migration, towards extending intelligent orchestration into new domains and applying AI to problems that were previously out of reach.