UiPath ScreenPlay is UiPath’s next‑generation “computer‑using” agent that automates user interfaces using natural language goals instead of brittle selectors. It brings agentic, adaptive behavior directly into Studio and Studio Web so automations survive UI changes and complex interactions.
What is UiPath ScreenPlay:
- UiPath ScreenPlay is an out‑of‑the‑box agent inside Studio that reads your natural‑language instructions, plans multi‑step UI actions, and executes them autonomously across applications.
- It is designed to navigate screens like a human, reading live UI context, understanding intent, and adapting when layouts, element positions, or themes change.
What are it's capabilities and features:
- Prompt‑to‑action UI automation: Converts plain language tasks (for example, “open last month’s invoice and download it”) into sequences of clicks, typing, selections, and scrolling, without manually configuring selectors.
- Resilient, context‑aware interaction: Uses large action models (LAMs), Computer Vision, DOM reading, and semantic understanding so automations keep working through UI drift, dynamic IDs, pop‑ups, iframes, and complex controls.
- Cross‑platform and rich actions: Supports Windows, Linux, and macOS, and exposes a rich set of UI Automation activities (get/set text, click/tap, swipe, select items, take screenshots, etc.) through the ScreenPlay activity family.
Integrations with the UiPath platform:
- UiPath Studio and Studio Web: ScreenPlay is delivered as part of UiPath.UIAutomation activities and appears as a dedicated ScreenPlay activity that you can drop into existing workflows.
- RPA, APIs, and agents: ScreenPlay is meant to complement, not replace, classic RPA and API calls, its UI intelligence can be composed inside larger processes orchestrated by Orchestrator or Maestro, alongside robots, APIs, and other agents.
- AI models: It can use UiPath‑hosted and third‑party large action models (for example, variants powered by Gemini, GPT‑4.1, Anthropic computer‑use) so you can tune speed, cost, and capability per use case.
Short example use case:
- A tester wants to validate hotel search results on a travel site: in Studio, they drop a ScreenPlay activity into a workflow and write “Open Hotels.com, search for 3‑star hotels in Paris next weekend, sort by price, and capture the top 5 results into a table.”
- ScreenPlay opens the browser, navigates, fills forms, handles pop‑ups, and extracts the data, while the rest of the workflow uses standard activities to save results to an Excel file and send a report email—creating a robust, low‑maintenance end‑to‑end automation.
Commentaires
Enregistrer un commentaire