Insurance claims teams live on SOPs: request a document, upload a photo, check coverage, issue a decision. For years the playbook was to build a bot that copies a human: move the mouse, click, fill fields, read the screen. But every time the policy system or mobile app updates, the bot breaks. You spend days remediating the selector, then two months later the next change hits. The backlog grows. The team burns out. Meanwhile the real work, making decisions, handling exceptions, reading unstructured attachments, still requires a human. The automation treadmill has stopped delivering value.
Why RPA breaks here
RPA tools like UiPath, Automation Anywhere, and Power Automate rely on selectors, XPath, and object IDs to target elements on a screen. If a new release shifts a field, adds a modal, or changes a class name, the bot halts or misfires. A 2023 industry survey found that 62 percent of RPA projects required rework within six months of a UI change. For insurance carriers that run hundreds of bots across policy systems, claims portals, and legacy mainframes, that means thousands of developer hours a year just to stay in place. The cost is not just labor. It is the risk of data entry errors, delayed claims, and compliance issues when bots land on the wrong field.
What changes with computer use agents
- Survives UI changes: agents see the screen and act relative to content rather than fixed selectors.
- No brittle selectors: they respond to text, labels, and visual context, so updates work without rebuilds.
- Recovers from exceptions: if a modal pops up or a form reloads, agents can read, adapt, and continue instead of crashing.
- Follows the SOP as written: a plain-English procedure directly becomes an agent prompt, no flowchart needed.
- Works on legacy and Citrix: agents run on any accessible desktop, including virtualized and thin-client environments where traditional RPA struggles.
RPA is great for high-volume, stable, backend tasks. Computer use agents are the durable way to automate the changing, exception-heavy front office.
How to move without the risk
You do not need to rip out everything tomorrow. Start with one high-pain process that is SOP-driven and exception-heavy, such as a first notice of loss intake or a document upload verification. Run a small pilot with a computer use agent in a cloud VM. Compare the build time, maintenance effort, and exception handling against your current RPA workflow. If the agent reduces rework from weeks to days, scale the approach to other claims stages. Reserve traditional RPA for the few stable, high-volume backend tasks where it still shines. This phased approach lets you build confidence while keeping risk low.
The next generation of claims automation is already here: agents that see, read, and adapt, not bots that break on every UI change. Talk to the Coasty team to see how computer use agents can replace the brittle RPA treadmill with a durable, SOP-driven workflow. Book a demo at https://cal.com/coasty/15min .
Want to see this in action?
View Case Studies