Every automation center of excellence has a backlog of projects that never ship. The reason is usually the same: the software that the bot needs to click or type into keeps changing. A corporate portal redesign, a new ERP version, or a minor UI tweak breaks the selector, and suddenly developers are on the rebuild treadmill. The bot halts, a human has to intervene, and the project slips another six months. In many enterprises, the cost of maintaining legacy RPA is equivalent to building new automation, and the backlog grows faster than the team can clear it.
Why RPA breaks here
Traditional RPA works by binding to specific UI elements, selectors, xpaths, object IDs. When an application changes a class name or moves a button, the bot sees nothing to click. The enterprise impact is measurable. Analysts estimate that around 30 to 40 percent of RPA maintenance effort goes toward fixing or rebuilding bots after UI changes. In large organizations, a single bot that should run autonomously can generate weeks of developer time every year. The real cost is not just the hours spent on rebuilds, but the missed opportunities to automate processes that stay in the backlog because the team cannot keep up with the maintenance burden.
What changes with computer use agents
- Survives UI changes without a rebuild
- No brittle selectors or object IDs required
- Recovers from exceptions instead of halting
- Follows the SOP as written, not a flowchart
- Works on legacy systems, Citrix, and virtualized desktops
Selectors are brittle. Seeing the screen is durable.
The durability advantage in practice
Computer use agents see the screen and act like a human. They move the mouse, click, type, and read the results. Because they operate at the user interface level, a change in a button location or a new field name does not break the agent. If an unexpected state appears, the agent can detect it and take an alternative path instead of stopping. This makes computer use agents suitable for processes that rely on standard operating procedures written in plain English. The SOP becomes a natural prompt. A human might write, 'Check the dashboard, click the download button, then upload the file to the legacy system.' The agent follows that instruction directly without needing a developer to build a flowchart or hardcode selectors.
How to move without the risk
Enterprise leaders can reduce risk by picking one high-pain process that is currently blocked or requiring manual intervention. Fast-moving processes with frequent UI updates are ideal candidates. Run a pilot with a computer use agent to compare uptime, exception handling, and maintenance effort against the existing RPA or manual approach. Measure metrics such as the percentage of time the bot works autonomously and the number of maintenance tickets logged. Use those results to build confidence across IT and operations. Expand the adoption to other processes once the business case is clear. This phased approach lets teams leverage the strengths of both RPA and computer use agents. RPA still fits very well for high-volume, stable, backend tasks where data is accessed via APIs. Computer use agents handle the long tail of changing UIs, exception-heavy workflows, and SOP-driven processes.
The durability of computer use agents comes down to seeing the screen instead of depending on brittle selectors. If your automation backlog is growing because bots break with every UI change, it is time to reconsider your automation strategy. Talk to the Coasty team to see how agents can replace brittle bots with durable automation. Book a demo at https://cal.com/coasty/15min .
Want to see this in action?
View Case Studies