The CIO case for moving from RPA to computer use agents
Every major bank we talk to has a backlog of RPA bots that need rebuilding. A single UI tweak in a core system can stop a bot in its tracks, forcing a developer to hunt down new selectors, test again, and redeploy. The cost is real. Gartner estimates that software maintenance can consume nearly half of total IT spend, and traditional RPA is a big piece of that. When the UI changes, the bot breaks and a developer has to rebuild it. That is the maintenance treadmill.
Why RPA breaks here
Traditional RPA tools like UiPath, Automation Anywhere, Blue Prism, and Power Automate rely on selectors, xpaths, and object IDs. These are brittle links to specific elements on a page. When an application updates its layout, those references can break, and the bot halts. You cannot simply patch it in place. You must rebuild the bot from scratch. The rebuild cost is not just hours of development. It is regression testing, change approvals, and the risk of new bugs. Analysts report that a significant percentage of RPA incidents stem from UI changes, not from logic errors. That means every upgrade in your ERP, CRM, or HR system creates new work for your automation team. The more apps you touch, the more you pay the maintenance tax.
What changes with computer use agents
- ●Survives UI changes: Agents see the screen like a person, so they find the right button or field even if labels or layout shift.
- ●No brittle selectors: No xpaths or object IDs to maintain. The agent reads the visual result and acts accordingly.
- ●Recovers from exceptions: When something unexpected happens, an agent can pause, reason about the state, and try another path instead of failing.
- ●Follows the SOP as written: A standard operating procedure in plain English is already almost a prompt. A computer use agent can run it directly, with no flowchart bot to build and babysit.
- ●Works on legacy and Citrix: Agents control the desktop, so they operate on systems where traditional RPA struggles, including virtualized environments.
RPA is great for high volume, stable, backend tasks. Computer use agents are the durable answer for changing UIs, exception-heavy workflows, and SOP-driven processes.
How to move without the risk
You do not need to rip and replace everything tomorrow. Start with a process that clearly demonstrates the pain of brittle selectors and frequent rebuilds. Choose a workflow where SOPs are already documented in plain language. Run a pilot with a computer use agent. Measure how long it takes to set up compared with rebuilding an RPA bot, and track how often it succeeds on its own. Use those results to build a business case for expanding the approach. Meanwhile, keep RPA for the work that is stable, high volume, and driven by APIs. Over time, you can gradually shift more SOP-driven tasks to computer use agents, reducing your maintenance backlog and freeing your automation team to focus on new opportunities.
The CIO case is now concrete: RPA bots are expensive to maintain when UIs change, and computer use agents are built to adapt. If you want to see how a computer use agent can run your high-pain SOPs without the rebuild cycle, book a demo with the Coasty team at https://cal.com/coasty/15min .