The finance and accounting close is one of the most labor-intensive cycles in any enterprise. Teams reconcile ledgers, populate journals, and validate figures across systems, often by copying numbers between spreadsheets, ERP screens, and email threads. When the process is documented in a standard operating procedure, the human steps still dominate. When it is automated with legacy RPA, the bots dominate but break whenever the UI changes or an exception occurs. The result is a recurring backlog of maintenance tickets and a close that remains dependent on manual overrides.
Why RPA breaks here
Traditional RPA for the close relies on selectors, XPaths, and object IDs to interact with screens. When the ERP interface is upgraded, when a business unit changes a field label, or when a workflow moves to a different version of the application, those selectors become invalid. The bot halts or produces wrong data. A developer must identify the change, re-create the selector, and redeploy the bot. In many organizations, the cost of maintaining a handful of close bots exceeds the value they deliver because every UI refresh renews the maintenance treadmill. Maintenance is not just time. A study of large enterprises shows that RPA projects often spend 30 to 40 percent of their total cost of ownership on maintenance, with more than half of that effort spent on fixing broken bots after changes. In high-volume, stable back-office processes, that cost can be justified. In the close, where workflows change each period and exception handling is unpredictable, the maintenance burden becomes a drag, not an asset.
What changes with computer use agents
- Survives UI changes. Agents read the screen and adapt without hard-coded selectors.
- No brittle selectors. Interaction is based on visual context and OCR, not fragile identifiers.
- Recovers from exceptions. When a step fails, the agent evaluates the current state and takes the next logical action instead of stopping.
- Follows the SOP as written. A plain-English procedure can be fed directly to an agent, so no flowchart or low-code designer is required.
- Works on legacy and Citrix. Agents see virtualized desktops and terminal emulations just as they see modern web UIs.
Selectors lock you into yesterday’s UI. Computer use agents see the screen and adapt to today’s reality.
How to move without the risk
You do not need to rip out all RPA at once. A pragmatic path starts with a single high-pain, SOP-rich process. For example, the month-end reconciliation of a key GL account across a couple of ERP forms and a spreadsheet. Identify a process that is documented in plain English, has clear hand-offs, and frequently surprises the team with unexpected states. Feed the procedure into a computer use agent in a pilot environment. Compare the time to complete the process, the number of maintenance tickets saved, and the reduction in manual overrides. Measure the results on a real close cycle. If the agent achieves consistent, repeatable outcomes with fewer incidents, expand to adjacent processes. Keep legacy RPA where it still makes sense, high-volume, deterministic batch jobs that do not change. Over time, replace the brittle bots in the close with agents that adapt to each month’s workflow and exception patterns. This phased approach lets you build experience with agents, validate their reliability, and preserve the value of existing automation.
The finance and accounting close does not have to be defined by recurring maintenance. Computer use agents can turn your SOPs into durable automation, surviving UI updates and handling exceptions without developer intervention. If you are ready to see how agents can simplify your close, book a demo with the Coasty team at https://cal.com/coasty/15min.
Want to see this in action?
View Case Studies