How to Pilot Computer Use Agents Alongside Your Existing RPA
You’ve built a center of excellence. You’ve automated the back-office work that never sleeps. But you still have a maintenance backlog. Every time IT changes a field name, a form moves, or a link shifts, your bots break. The team spends more time rebuilding bots than automating new work. Some processes are still stuck in the queue because they live on legacy systems, Citrix, or custom portals where traditional RPA cannot see anything but a screen capture. The SOPs exist, but only a human can follow them step by step. You need a way to add automation that does not force you to rewrite everything when the tech changes.
Why RPA breaks here
Traditional RPA (UiPath, Automation Anywhere, Blue Prism, Power Automate) binds to selectors, XPath patterns, and object IDs. These identifiers are brittle. When an application updates a field name, moves a button, or changes its DOM structure, the bot stops. You must send a developer to rebuild it. The cost is not just the hours of coding. It is the downtime, the risk of missed deadlines, and the fatigue on a team that is already stretched thin. Industry benchmarks show that up to 40 percent of automation hours can go into maintenance rather than new development. For processes that run on legacy systems, Citrix, or custom-built portals, that number is even higher. The bots halt on exceptions. They do not look at the screen to understand what went wrong. They simply throw an error and wait for a human to intervene. This makes them unsuitable for high-variability work where the user experience changes often.
What changes with computer use agents
- ●Survives UI changes without rebuilding the bot
- ●No brittle selectors or object IDs to maintain
- ●Recovers from exceptions by reading the screen and adapting
- ●Follows the SOP written in plain English
- ●Works on legacy applications and Citrix where RPA struggles
- ●Controls desktops, browsers, and terminals directly, not just APIs
RPA binds to the UI. Computer use agents act like a human: they see the screen, read the text, and click accordingly. That is why they survive change instead of breaking it.
How to move without the risk
You do not need to rip out everything tomorrow. Start with a single, high-pain process that has three characteristics. First, it runs on a changing interface. Second, it has an existing SOP that explains the steps in plain language. Third, it is important enough to justify a pilot but not so critical that a failure would cause a business outage. Pick a process like employee onboarding for contractors, a compliance data collection task, or an expense report submission that lives on a legacy portal. Use the SOP as the prompt for a computer use agent. Run it in parallel with your current RPA bot. Compare uptime, maintenance time, and exception handling side by side. If the agent handles the UI updates without breaking, you have found your first durable addition to your automation stack. After the pilot, expand to other processes that fit the same pattern. Keep the RPA bots for high-volume, stable, backend tasks where they excel. Over time, you will replace the brittle workflows with agents that can adapt on their own. This phased approach lets you move forward without betting the whole house on a new technology.
Computer use agents give you the durability you need to stop rebuilding bots after every UI change. They follow SOPs, recover from exceptions, and work where traditional RPA cannot. Ready to see how agents can complement your existing RPA? Book a demo with the Coasty team at https://cal.com/coasty/15min .