Your RPA center of excellence has spent years automating the low-hanging fruit. Now you face a different problem: processes with changing interfaces, frequent errors, or strict operating procedures that are too complex for a flowchart bot to follow. When the finance portal redesigns or HR adds a new workflow step, your bots break and a developer must spend hours rebuilding them. That is the maintenance treadmill. It is also why many enterprises are now building an AI agent center of excellence to handle the long tail and exception-heavy work that traditional RPA cannot survive.
Why RPA breaks here
Traditional RPA tools such as UiPath, Automation Anywhere, and Blue Prism rely on selectors, xpaths, and object IDs to find controls on a screen. When a developer builds a bot, they assume a stable UI. If the application updates its CSS classes, changes a navigation tree, or swaps a button label, the selector becomes invalid and the bot halts. The bot does not recover. A developer must open the process, adjust the selector, re-test, and redeploy. This rebuild-on-change cycle is the single biggest contributor to RPA maintenance cost. Industry data shows that enterprises spend roughly 30 to 40 percent of their RPA budget on maintenance and rework, not on new automation. In many organizations, that maintenance backlog grows faster than the team can handle, so only the most stable processes get automated and the rest remain manual.
What changes with computer use agents
Computer use agents control desktops, browsers, and terminals by seeing the screen and acting like a human. They move the mouse, click, type, and read the result. This fundamental difference changes three things that matter to an automation center of excellence.
- Survives UI changes: agents see the current state of the UI instead of relying on fixed selectors. When a button moves or a field changes, the agent recalculates its target on the fly.
- No brittle selectors: developers do not need to maintain object repositories. The agent uses visual cues and context to locate controls across applications.
- Recovers from exceptions: when a bot hits an unexpected error, a computer use agent can look at the screen, decide on a corrective action, and continue. RPA bots typically halt and require developer intervention.
- Follows the SOP as written: a standard operating procedure written in plain English is already almost a prompt. A computer use agent can follow it directly without a flowchart bot to build and babysit.
- Works on legacy and Citrix: many enterprises still run processes on virtualized or terminal-based environments where RPA struggles. Computer use agents can interact with these sessions the way a human user does.
A computer use agent sees the screen and adapts. That is the core difference that turns a brittle bot into a durable automation asset.
How to move without the risk
You do not need to replace every RPA process overnight. A pragmatic path to a durable automation program looks like this. First, identify one high-pain process that is manual or brittle RPA. It should have frequent UI changes, many exception cases, or a detailed SOP that is hard to encode in a flowchart. Second, pilot a computer use agent on that process. Compare the time it took to implement and maintain the RPA version versus the agent. Measure exception rates, uptime, and the time that humans spend on handoffs. Third, expand to additional processes that share the same characteristics. As the team gains confidence, you can gradually shift more exception-heavy and SOP-driven work to agents. You can also continue to run high-volume, stable, backend operations on RPA where it still fits well. The goal is to build an AI agent center of excellence that handles the changing and complex work, while your existing RPA investments continue to serve the predictable, high-volume tasks.
Why agents fit the long tail of automation
Enterprise processes are rarely as stable as the V1 of a modern SaaS platform. Business units change workflows to meet new regulations, customers, or internal policies. IT departments roll out updates that improve accessibility or performance. Agents that can adapt to these changes reduce the need for constant developer intervention. They also support processes that are defined by documentation: a procurement request, an onboarding checklist, or an incident response procedure. These are exactly the processes where a standard operating procedure already exists. A computer use agent can follow the same steps as a human, with the advantage of being able to read the screen and react to what it sees. This combination of adaptability and documentation-driven automation is what makes agents a durable foundation for an automation center of excellence.
RPA will continue to play a role in high-volume, stable workflows. The real opportunity for an automation center of excellence is to build a durable layer of computer use agents for the processes that change, break, and require human-like judgment. To see how agents can handle your exception-heavy and SOP-driven work, book a demo with the Coasty team.
Want to see this in action?
View Case Studies