Your SOPs Are Already the Prompt: Automating Them with AI Agents
Your finance team has a two-page runbook for reconciling vendor invoices. A manager writes it in plain English, the team follows it every month, and the work is done. But if you hand that same text to an RPA developer, they usually say "I need screenshots and a flowchart." The bot breaks in the next UI upgrade and you are back at the drawing board. That is the maintenance treadmill. Your SOPs already contain almost everything an intelligent agent needs, but most RPA stacks never use them.
Why RPA breaks here
Traditional RPA tools like UiPath, Automation Anywhere, Blue Prism, and Power Automate rely on selectors, xpaths, and object IDs to find UI elements. When a UI refreshes, a new class name is added, a layout shifts, or a screen is rendered dynamically, the bot hits a broken assertion and halts. A developer must open the bot, inspect the new selector, patch it, and redeploy. Many organizations track an average of three to eight bot rebuilds per year for every critical process, often driven by single UI updates. The cost compounds when the process spans multiple applications. A single business change can ripple through ten bots, each needing a rework. The result is a backlog of "fix this bot" tickets that grow faster than the automation team can close them.
What changes with computer use agents
- ●Agents see the screen and act like a human: they move the mouse, click, type, and read the result. They do not need brittle selectors or xpaths.
- ●When the UI changes, the agent notices the new layout, locates the same function by its label or context, and continues.
- ●Agents recover from exceptions instead of halting. If a popup appears or a field is disabled, they read the message, decide a next step, and retry.
- ●They can follow a standard operating procedure written in plain English. No flowchart or logic designer is required.
- ●Because they use vision and mouse/keyboard simulation, agents work across any app, including legacy systems, Citrix, and virtualized desktops where RPA struggles.
A computer use agent reads your SOP and follows it. When the UI changes, it sees the change and keeps going.
How to move without the risk
You do not have to retire all RPA overnight. Start with one high-pain process where the SOP is already documented and the UI changes frequently. Use Coasty to pilot the process with an agent that reads the SOP and runs on real desktops and browsers. Measure how many exceptions it hits, how many selector rebuilds it avoids, and how quickly the process completes. Compare that to the current RPA or manual effort. If the pilot shows consistent uptime and lower maintenance, expand to other SOP-driven workflows. Keep RPA for high-volume, stable, backend tasks where APIs exist and the UI does not change. Over time, you can transition more processes to agents, reducing the rebuild burden and freeing automation teams to work on higher-value initiatives.
Your SOPs are already almost a prompt for an intelligent agent. The next step is to see how a computer use agent can run a process end-to-end. Book a demo with the Coasty team to test one of your processes and understand what is possible without the RPA rebuild treadmill. Talk to the Coasty team at https://cal.com/coasty/15min .