Your mainframe and legacy applications process hundreds of thousands of transactions a month, but your bot backlog is growing. Every UI patch, every new screen, and every obscure error forces a developer to rebuild the bot from scratch. Meanwhile, the team that wrote the original standard operating procedure has moved on, and human runbooks are the only thing that actually works.
Why RPA breaks here
Traditional RPA relies on stable selectors, XPath rules, and object IDs to drive any application. When a mainframe emulator, a green-screen terminal, or a virtualized desktop changes a single character of the HTML or a class name on the screen, the bot fails. Gartner estimates that about 30% of RPA licenses sit idle because bots cannot run on legacy or virtualized environments. The cost of maintenance compounds quickly. Each bot that touches a changing UI typically requires a developer hour every two to three weeks just to keep it alive. Over three years, that can exceed the original license cost in pure rebuild and rework. When the process includes conditional logic, error handling, or hand-offs between systems, the backlog grows faster than the team can resolve it. The result is a shrinking pool of usable bots and a growing number of workflows that still require human execution.
What changes with computer use agents
- Survives UI changes without rebuilding the bot
- No brittle selectors or hard‑coded XPaths
- Recovers from exceptions and unexpected states
- Follows the SOP as written, not a flowchart
- Works on legacy terminals, Citrix, and virtualized desktops
Traditional RPA breaks on change. Computer use agents adapt to what they see instead of what they expect.
How to move without the risk
You do not need to rip out all your existing automation overnight. A pragmatic path forward starts with a single process that is high‑pain, high‑value, and bottlenecked by brittle bots or manual execution. Document the current SOP in plain language, then run it through a computer use agent in a pilot environment. Compare the time to complete the process, the rate of failures, and the total effort required to maintain each approach. If you already have an RPA footprint, consider keeping it on stable, backend tasks, payroll, basic invoice processing, and high‑volume, deterministic data entry, while computers use agents take over the changing front‑end workflows. Over time, you can expand the agent footprint to more processes, retire the most brittle bots, and build a more resilient automation stack.
Moving to a more durable automation model for mainframe and legacy applications starts with one process that proves the difference. Talk to the Coasty team to see how computer use agents can follow your SOP, recover from errors, and work across legacy environments without the rebuild‑on‑change treadmill. Book a demo at https://cal.com/coasty/15min.
Want to see this in action?
View Case Studies