A manufacturing plant runs on SAP S/4HANA or Oracle Fusion Cloud, with thousands of material moves, purchase orders, and work orders entered daily. A ten-person operations team manually types data from shop floor paper into the ERP. They copy-paste from spreadsheets, correct formatting, and chase missing values from operators. Errors slip through, inventory gets misstated, and production planners spend hours fixing downstream reports. To fix this, IT builds a UiPath bot. It types into the correct fields, but the moment the ERP updates its layout, the bot breaks. A developer must rebuild the selector-based workflow, test it, and redeploy. This cycle repeats every two to three months. The backlog of broken bots grows. The team spends more time rebuilding than automating. The real bottleneck is not the ERP. It is the maintenance treadmill of brittle RPA.
Why RPA breaks here
Modern ERPs update their screens every release. They add new columns, reorder fields, introduce modal dialogs, and change class names for buttons. Traditional RPA relies on selectors, xpaths, and object IDs. When the UI changes, those selectors fail. The bot halts or types into the wrong field. A study of RPA deployments in enterprises estimates that 30 to 40 percent of bot runtime is spent on maintenance and rework. Teams rebuild one bot every two to four months on average. For high-volume, stable backend tasks, this cost is acceptable. For front-office, change-heavy processes like ERP data entry, the cost is unsustainable. In manufacturing, operators move quickly. They change machines, routes, and material codes. A bot that halts on a missing field forces a human to intervene, creating new cycle time and risk. The process becomes a human-in-the-loop service, not automation.
What changes with computer use agents
- Survives UI changes
- No brittle selectors
- Recovers from exceptions
- Follows the SOP as written
- Works on legacy and Citrix
Computer use agents control the desktop like a human, move the mouse, click, type, read the result, so they survive UI and app updates and recover from exceptions instead of halting.
How to move without the risk
You do not need to rewrite every bot overnight. Start with one high-pain process, such as raw material receipt data entry. Document the current SOP in plain English, then pair it with the Coasty agent to validate that the agent can follow the steps. Run the agent on a pilot set of transactions. Measure cycle time, error rate, and maintenance time. Compare against the UiPath bot if one exists. If the agent achieves similar or better results with fewer rebuilds, expand to related processes. Over time, move more SOP-driven workflows from manual work to the agent. Keep the stable, high-volume backend tasks on RPA. Keep the changing, exception-heavy work on agents. This hybrid approach lets you capitalize on the durability of computer use agents while still using RPA where it fits best.
The next step is to see how a computer use agent handles your current ERP data entry process. Book a demo with the Coasty team at https://cal.com/coasty/15min .
Want to see this in action?
View Case Studies