RPA with Autopilot: what RPA is good for, how to brief a workflow, the role of UI Automation, and where human review matters most.

Robotic Process Automation (RPA) automates repetitive, rule-based work across the systems people use every day. With Autopilot, you describe the work in plain language and it produces a runnable workflow. Your part is to describe it well and review what comes back.

RPA is best for well-defined, deterministic work, especially where systems have no usable API:

Building a new workflow is only the starting point. Autopilot works across the whole life of an automation, so common requests include:

Autopilot produces more accurate output when you describe the outcome you want rather than listing activities. Provide the business context — what the process does, what triggers it, and what success looks like — and Autopilot determines the implementation.

A good brief includes:

For general prompting guidance, refer to Writing effective prompts. An existing process document is a strong starting point, as long as real customer data is removed first.

A UiPath RPA workflow can be built as a low-code workflow or as a coded one, and you guide which:

Both can also be mixed in one project. Choosing the shape up front, rather than mid-build, avoids a restructure later; without direction, Autopilot defaults to low-code.

The hardest and most valuable RPA work is driving an application's interface the way a person would — by clicking, typing, reading values, and moving through screens — for systems that offer no API. UiPath's UI Automation is what makes this reliable, and Autopilot uses it directly.

Rather than guessing how to find each element, Autopilot explores the application at design time: it captures the screen, identifies the elements to act on, and builds the steps to navigate and fill them. Captured elements are stored as stable targets in an Object Repository, so the same button or field can be reused across workflows and maintained in one place.

Interfaces change, and UI Automation is built to absorb that:

If the target application is not available to capture against, Autopilot still produces the real workflow, leaving each interaction marked for a developer to finish the capture in Studio. Generated selectors are best treated as drafts either way, and confirmed against the real application.

UI automation is also the most expensive kind of RPA to build. It takes more time and effort than file, data, or API work, so an API is the better path wherever one exists.

RPA projects have house rules worth stating once, rather than repeating every session. Common examples:

You can capture these conventions in an AGENTS.md file at the project root. Autopilot reads and follows the instructions defined there — folder naming rules, architecture decisions, coding standards — for the duration of the session.

Autopilot is strong at greenfield, well-structured RPA and at the first version of a workflow. It is weaker, and needs closer review, on:

Starting with Studio Desktop STS build 2026.0.197, Autopilot generates and edits RPA workflows through an iterative reasoning loop. On each request, Autopilot works in three phases:

Because Autopilot reads your project during the Discovery phase, it can reference and reuse workflows already in your project. Autopilot is aware of your current file, selected activity, and variable definitions — you do not need to explain your project structure before asking a question or making a request.

Context is also maintained across messages in the same session: a follow-up request such as "undo the last change" applies to the most recent operation, not the entire workflow.

For complex, multi-step tasks, allow Autopilot to complete its execution cycle before sending a follow-up message. Interrupting mid-task may leave the project in an intermediate state.

Autopilot leverages a range of tools to support your automation work. By default, all tools are enabled. Autopilot can:

Autopilot automatically loads relevant skills from the UiPath skills catalog based on the task context. Skills are instruction bundles that teach Autopilot how to perform specific UiPath tasks using platform conventions.

For the full list of skills, refer to the UiPath skills repository.