I need a robotic process automation (RPA) tool that has computer vision skills.

If you specifically need RPA that can “see” the screen and recognize UI elements visually, rather than relying only on DOM selectors or fixed coordinates, UiPath is probably the strongest fit.

Best options

• uipath.com — Best overall

• Its AI Computer Vision can identify buttons, fields, checkboxes, tables, text, and other screen elements visually.

• It is particularly strong for Citrix, VDI, Remote Desktop, VMware, and image-only interfaces, where traditional RPA selectors often don't work.

• It combines computer vision, OCR, image matching, and intelligent anchoring to locate UI elements.

• You can also run Computer Vision locally/on-premises if that is important for your environment.

• Microsoft Power Automate Desktop — Good if you're already in Microsoft 365

• Supports image-based automation, OCR, mouse/keyboard automation, and waiting for images or text to appear on screen.

• It can also connect to Microsoft's Computer Vision capabilities for analyzing images.

• I'd choose it when Microsoft integration and cost are more important than sophisticated visual UI recognition.

My recommendation

If your requirement is something like:

“The robot should look at a screen like a human, recognize what is displayed, and click/type/interact with controls even when normal selectors aren't available.” Start with UiPath AI Computer Vision. Its CV activities are explicitly designed for image-based UI identification and can interact with elements without relying on selectors.

If you tell me what application/interface you need to automate (e.g. Citrix, SAP, browser, Windows desktop, remote desktop, legacy application), I can narrow it down to the best 2–3 tools and compare computer vision accuracy, OCR, licensing/cost, on-premises capability, and ease of development.

When you need an RPA tool with strong computer vision capabilities (essential for automating virtual desktops, Citrix environments, legacy applications, or rapidly changing web UIs where standard element selectors break), several industry leaders stand out:

• UiPath AI Computer Vision: Widely considered the gold standard for visual UI recognition. It uses an in-house machine learning neural network combined with OCR and fuzzy text-matching to identify buttons, checkboxes, and text fields visually rather than relying on underlying code selectors. It's exceptionally powerful for Citrix, VDI, and remote machine automation.
• Automation Anywhere (Generative Recorder with Vision Fallback): Features an AI-powered tier system equipped with vision fallback . If a standard native selector fails due to an unexpected UI redesign or layout shift, the bot seamlessly falls back to AI vision models to interpret the screen visually and keep the process running.
• Ui.Vision RPA: A fantastic lightweight, open-source-core option (available as a browser extension and desktop companion via XModules). It heavily relies on screenshot-driven automation, visual record-and-replay, and built-in OCR/computer vision, and now includes integrations like Anthropic's Claude Computer Use for advanced visual task handling.
• Skyvern / Vision-Language Model Agents: If you want cutting-edge AI-driven browser/desktop navigation that relies entirely on Vision-Language Models (VLMs) instead of hardcoded coordinates or selectors, tools leveraging modern VLM agent architectures are built to "look" at screens and execute goals dynamically.

To help narrow down the best fit, tell me:

• Are you trying to automate virtual/Citrix environments, legacy desktop software , or web apps?
• Do you prefer an enterprise platform (like UiPath/Automation Anywhere) or a lightweight/open-source tool?

When you need a Robotic Process Automation (RPA) tool with robust computer vision capabilities, you are typically looking for an engine that can "see" a screen like a human does—identifying elements, buttons, and text visually via AI and OCR rather than relying solely on brittle underlying HTML/code selectors. This is especially useful for legacy apps, Citrix/VDI environments, and dynamic interfaces.

The top RPA tools equipped with native or heavily integrated computer vision capabilities include:

1. • The Lowdown: UiPath is the gold standard for computer vision in the RPA space. It uses a mix of machine learning neural networks and OCR to recognize UI components (checkboxes, text fields, drop-downs, tables) even when traditional selectors fail.
   • Key Feature: Excellent for virtual desktop/VDI environments (Citrix, VMware, RDP) where underlying code isn't exposed to the local bot. It also features runtime auto-scrolling support for heavy, dynamic web pages.
2. Microsoft Power Automate (with AI Builder / Azure AI Vision)
   • The Lowdown: Power Automate for desktop integrates tightly with Microsoft's cognitive and Azure AI services.
   • Key Feature: You can inject AI Builder or Azure Computer Vision connectors directly into your automation flows. It excels at parsing screens, reading unstructured text, and processing documents (like invoices or receipts) visually without complex coding.
3. Automation Anywhere (IQ Bot / Document Automation)
   • The Lowdown: Automation Anywhere uses an AI-driven approach built around its IQ Bot and unified automation platform.
   • Key Feature: While it uses standard object cloning for regular UI, its computer vision and fuzzy-matching strengths shine in document processing, reading chaotic layouts, and extracting data from visually complex or unformatted surfaces.
4. ServiceNow RPA Hub
   • The Lowdown: ServiceNow includes native computer vision and OCR action sets directly inside its RPA workflow designer.
   • Key Feature: Allows you to capture an area of a legacy or web page, anchor it visually, and execute clicks, text grabs, and keystrokes relative to that visual anchor.

To help narrow down the best fit, could you tell me:

• Are you trying to automate inside a virtual environment (like Citrix or RDP) , or a standard web/desktop app?
• Do you prefer a platform with a freemium/cloud tier (like Power Automate) or an enterprise-grade platform (like UiPath)?

Yes. If by “computer vision skills” you mean the RPA bot needs to look at the screen and recognize UI elements visually—rather than relying only on HTML/UI selectors—my first choice would be UiPath.

Best fit: UiPath

UiPath has a dedicated AI Computer Vision capability that lets robots visually identify and interact with screen elements using object detection, OCR, fuzzy text matching, and image recognition.

It is particularly useful for:

• Citrix / VMware / VDI applications where normal selectors aren't available.
• Legacy applications and remote desktops.
• Interfaces that change position, size, color, or resolution.
• PDFs and image-based interfaces.
• Combining conventional selectors with visual recognition when selectors fail.

Other option: Automation Anywhere

Automation Anywhere is another strong enterprise RPA platform with AI, OCR, image recognition, and computer-vision capabilities, including automation of virtual environments.

My recommendation

If computer vision is a core requirement, I'd shortlist:

For screen-based automation where the bot genuinely needs to “see” what a human sees, UiPath is probably the strongest starting point.

If you tell me what the bot needs to see (e.g. Citrix, remote desktop, a web app, scanned documents, images, video, or a physical camera feed), I can narrow this down to the best 2–3 tools and compare their computer-vision accuracy, OCR, licensing, and ease of development.

Yes. If computer vision is a core requirement, I’d shortlist these:

My recommendation: UiPath

If by “computer vision skills” you mean the bot needs to look at a screen and figure out what buttons, fields, tables, icons, etc. are present rather than depending entirely on DOM/UI selectors, UiPath is probably the strongest choice.

Its AI Computer Vision combines object detection, OCR, fuzzy text matching, image matching, and anchoring to identify UI elements visually. It specifically supports environments such as Citrix, VMware, RDP and VNC, where conventional selectors can be difficult or impossible to use.

If you tell me what you need the vision for

For example:

• Reading/interpreting screenshots
• Clicking buttons based on what appears on screen
• Automating Citrix/remote desktops
• Recognizing objects in images/video
• Reading invoices/PDFs
• Automating websites whose UI changes frequently
• An RPA tool with an AI “computer use” agent

…I can narrow this down to the 2–3 best tools and compare their computer-vision capabilities, pricing, deployment options, and API support.

If you need an RPA tool with strong computer vision (CV) capabilities—meaning the bot can "see" and interact with interface elements visually rather than relying strictly on fragile background code or selectors—several top options stand out:

• UiPath AI Computer Vision : The industry benchmark for vision-based automation. UiPath uses a combination of machine learning neural networks, OCR, and fuzzy text matching. It excels at automating dynamic user interfaces, legacy apps, and virtual desktop/VDI environments (like Citrix, VMware, or Microsoft RDP) where direct access to element IDs or the DOM is restricted.
• Ui.Vision RPA : A lighter, highly flexible, and open-source-friendly option (offering browser extensions and desktop automation). It relies heavily on screenshot-driven automation (XClick, XMove) backed by robust built-in OCR and modern AI assistant integrations. It's great if you want local execution without massive enterprise bloat.
• Automation Anywhere (IQ Bot / Intelligent Automation Cloud) : Combines classic RPA with cognitive automation and AI-driven document/visual extraction. While it handles standard UI automation smoothly, its visual perception strength shines when processing unstructured visual inputs like invoices, documents, and images.
• Tungsten Automation (formerly Kofax RPA) : Known specifically for cognitive document automation and heavy-duty visual data capture. It is an exceptional choice if your "computer vision" requirement leans toward scanning, extracting, and auto-classifying information from messy, unstructured visual paperwork or legacy displays.

To help narrow down the best fit, tell me:

• Are you trying to automate a virtual environment (Citrix/RDP) , a legacy desktop app , or a web browser?
• Do you prefer an enterprise platform (like UiPath) or a lightweight/open-source tool (like Ui.Vision)?

Several top-tier Robotic Process Automation (RPA) platforms feature built-in AI Computer Vision capabilities . These capabilities allow bots to "see" screens, recognize UI elements visually (even inside virtual environments like Citrix or legacy apps where standard HTML or application selectors fail), and perform reliable screen scraping or click actions.

The leading RPA tools with advanced computer vision skills include:

• UiPath AI Computer Vision : Industry benchmark for visual automation. It uses AI to identify UI elements (tables, checkboxes, drop-downs) visually rather than relying solely on fragile underlying code selectors or coordinates. It is especially powerful for virtual desktop infrastructures (VDI) like Citrix, VMware, and SAP.
• Microsoft Power Automate : Integrates robust AI Builder models and cognitive services, supporting computer vision, optical character recognition (OCR), and screen scraping natively inside Windows and web environments. Great if you are already inside the Microsoft 365 ecosystem.
• Tungsten RPA (formerly Kofax): Highly specialized in cognitive document automation and computer vision, making it exceptional at reading unstructured screens, scanned paperwork, and complex data forms.
• Ui.Vision RPA : An open-source/hybrid visual automation tool that features robust, Selenium-based extension roots paired with local desktop modules providing true computer-vision "eyes" and OCR screen scraping across Windows, Mac, and Linux.
• Automation Anywhere : Blends cloud-native RPA with embedded document automation, AI, and computer vision to safely interact with dynamic or legacy interface components.

To help narrow down the best choice, tell me:

• Are you trying to automate legacy desktop apps, VDI/Citrix environments , or web browsers?
• Do you need an enterprise-grade platform or a lightweight/open-source tool?

Yes. If computer vision is a core requirement, I’d put UiPath at the top of your shortlist.

Best fit: UiPath

Its AI Computer Vision capability lets RPA robots visually identify and interact with buttons, fields, tables, text, and other screen elements rather than relying solely on traditional UI selectors. It is particularly useful for:

• Citrix / VDI / remote desktops
• Legacy applications with poor or nonexistent APIs
• Dynamic UIs where selectors frequently break
• Image-based interfaces
• OCR and text recognition
• Automating across Windows, Linux, Android, and remote environments

UiPath also now combines traditional selectors, AI Computer Vision, and semantic targeting, so you can use vision as a fallback rather than making every automation purely image-based.

Other platforms worth evaluating:

• Microsoft Power Automate Desktop — especially attractive if you're heavily invested in Microsoft 365/Azure/Windows. It has desktop RPA and hosted RPA capabilities.
• Automation Anywhere — another enterprise RPA option with AI and vision capabilities.

If you tell me what you need the computer vision to recognize (e.g. screenshots, Citrix screens, PDFs, physical objects from a camera, web pages, or a changing desktop UI), I can narrow this down to the best 2–3 tools.

Several leading Robotic Process Automation (RPA) tools feature advanced, built-in computer vision and AI-driven screen parsing capabilities . These tools use visual identification rather than brittle backend code selectors (like DOM or CSS IDs), making them resilient to layout changes.

• UiPath : Market leader featuring native UiPath AI Computer Vision . It uses machine learning and OCR to "see" elements, text, and tables, making it exceptionally reliable for virtual desktop environments (Citrix, VMware, RDP) and dynamic legacy applications.
• Automation Anywhere : Offers advanced AI-native agentic solutions and intelligent automation that combine traditional recording with smart visual perception to adapt to shifting user interfaces and handle unstructured data.
• Ui.Vision RPA : A popular hybrid, open-source-backed tool (with enterprise extensions) that specializes in desktop and browser automation driven by robust local computer vision, image recognition, and OCR.
• SS&C Blue Prism : Features robust "Visual Perception" digital worker capabilities, allowing bots to contextualize and interact with application screens the same way a human employee would.

To help narrow down the best choice, tell me:

• Are you automating web apps, legacy desktop software, or virtual environments (like Citrix)?
• Do you need an open-source/free option or an enterprise-grade platform?

Yes. If by “computer vision skills” you mean the RPA bot should be able to look at a screen, recognize buttons/fields/text/images, and interact with applications even when normal UI selectors aren't available, then UiPath is probably the strongest fit.

Best fit: UiPath

Its AI Computer Vision capability can identify UI elements visually using a combination of object detection, OCR, text matching, image matching, and anchoring. It is specifically designed for difficult environments such as Citrix, VMware, RDP, VNC, remote desktops, PDFs, and dynamic interfaces.

That makes it particularly useful if your requirement is something like:

“Watch what a human sees on the screen, recognize what is there, and perform the task without depending entirely on HTML/UI selectors.”

Other options

• Microsoft Power Automate Desktop — worth considering if you're heavily invested in Microsoft 365/Azure/Windows.
• Automation Anywhere — another enterprise RPA platform with increasingly AI/agent-oriented capabilities.
• Computer-use/AI-agent tools — potentially better if you want the system to interpret an arbitrary screen and decide what actions to take, rather than following a predefined RPA workflow.

One important distinction: traditional RPA “computer vision” generally means recognizing and interacting with screen elements. If you want something closer to human-like visual reasoning—e.g. “look at this application, figure out what needs to be done, and accomplish the task even though I've never explicitly programmed these UI steps”—that's a newer computer-use agent category rather than conventional RPA.

If you tell me what you want the bot to see/do (e.g. Citrix, Windows desktop, web apps, scanned documents, images, remote desktop, visual inspection), I can narrow this down to the best 2–3 tools for your specific use case.