What Happens to Your RPA Developers When AI Agents Take Over
Your automation center of excellence has a solid roster of RPA bots. But in the last six months, three of them broke because the vendor released a patch that shifted a button two pixels to the right. Your developers spent days rebuilding selectors, finding new xpaths, and retesting. Meanwhile, the SOP for that process still reads like a human instruction manual, but no human can execute it anymore. That is the cost of staying on RPA in a changing world.
Why RPA breaks here
Traditional RPA binds tightly to selectors, xpaths, and object IDs. When the application changes layout, version, or even a font, the bot halts. Industry research suggests that one in every five RPA bots requires a rebuild within months. A midsize bank we benchmarked spent 32 percent of its automation budget on maintenance, not on new processes. That is the maintenance treadmill. Your developers are constantly putting out fires instead of building new value. The problem is not the automation itself. It is the fragile connection between the bot and the UI.
What changes with computer use agents
- ●Survives UI changes: agents see the screen and act like a human. When a button moves, they find it again.
- ●No brittle selectors: they do not rely on xpaths or object IDs. They use visual context instead.
- ●Recovers from exceptions: when something unexpected happens, they read the error screen, adjust, and continue.
- ●Follows the SOP as written: a process documented in plain English can be executed directly by the agent.
- ●Works on legacy and Citrix: they interact with virtual desktops and systems that RPA cannot touch.
The one line a VP of automation should remember: selectors are the weak link; computer use agents make automation durable.
How to move without the risk
Do not rip and replace everything at once. Start with one process that has high maintenance cost and frequent UI changes. Document it as a clear SOP. Run a pilot with a computer use agent to see how it handles real desktops, browsers, and terminals. Measure the time saved on maintenance versus the cost of the pilot. When you are confident, expand to other processes. Keep RPA for high-volume, stable, backend tasks. Use computer use agents for the long tail and exception-heavy work. This phased approach lets you hedge your bets while building momentum toward a more durable automation strategy.
The future of automation is not about choosing between RPA and AI. It is about using the right tool for the right job. To see how computer use agents can replace brittle bots in your environment, book a demo with the Coasty team at https://cal.com/coasty/15min .