Most large enterprises have run unattended RPA bots for years. They claim 24/7 coverage, but the reality is often a small team of engineers constantly babysitting a fragile fleet. A UI change, a new version of a web portal, or a slight layout shift causes the bot to fail. The team wakes up at 3am to fix selectors or tweak workflows, and the cycle repeats every time something changes. This is the invisible cost of legacy RPA: high operational risk, unpredictable downtime, and a growing backlog of repair tickets.
Why RPA breaks here
Legacy RPA works by binding to specific UI elements. Developers write selectors, xpaths, and object IDs that point to buttons, tables, and input fields. When the application updates its styling or layout, those references can break. You then have to capture new selectors, rebuild the bot, and redeploy. Gartner and other analysts estimate that a substantial portion of RPA total cost of ownership comes from maintenance and exception handling rather than initial development. For many teams, the majority of bot tickets are not new features but bug fixes for broken selectors. This creates a treadmill where the value of automation is constantly eaten by the effort to keep bots running.
What changes with computer use agents
- survives UI changes
- no brittle selectors
- recovers from exceptions
- follows the SOP as written
- works on legacy and Citrix
Computer use agents replace brittle selectors with visual perception: they see the screen and act like a human. When the UI changes, the agent notices the new layout and continues. When an exception occurs, it reads the error and works around it instead of halting.
The on-call impact
An RPA bot that halts on the first exception means an alert, a pager, and a person reviewing logs. A computer use agent that can recover from exceptions or ask for clarification stays online. The bot handles more of the variability that historically required human intervention. This reduces the frequency of on-call escalations and frees your team to focus on new automation opportunities instead of triage. The result is a more reliable, lower-maintenance automation strategy.
How to move without the risk
You do not have to rip out every legacy bot at once. Start with one high-pain process: something that fails often, has complex layouts, or runs on systems that RPA struggles with like Citrix or legacy web apps. Run the existing RPA bot alongside a computer use agent for the same task. Compare uptime, exception rates, and maintenance effort. If the agent is more reliable and requires less hand-holding, expand it to other processes. Over time, rework older bots that are too complex or brittle to maintain. Legacy RPA still works well for stable, high-volume, backend tasks. The win for computer use agents is the long tail of processes that change, require judgment, or live on systems that resist traditional automation.
Why this matters for automation leaders
As your automation footprint grows, the cost of maintaining brittle bots grows faster than the number of bots. A durable automation strategy should minimize on-call risk and maximize resilience. Computer use agents let you automate processes that were previously too risky or too variable to automate with traditional RPA. They reduce the number of alerts that reach your team and let you run more unattended bots with confidence.
If your automation strategy means waking up at 3am to fix broken bots, it is time to rethink your approach. Computer use agents see and adapt instead of breaking. Book a demo with the Coasty team to see how they can reduce your on-call burden and build a more durable digital workforce.
Want to see this in action?
View Case Studies