Platform-native Automation Threatens RPA

The real threat is that automation is moving from separate bot software into the software people already use every day. Automation Anywhere built its business on teaching bots to click through screens, move data between systems, and read documents across messy enterprise stacks. As Microsoft and Salesforce add natural language agents and workflow actions inside their own products, some of that work can happen at the source instead of through an external RPA layer.

• Automation Anywhere is strongest where work spans many old and new systems. It sells yearly contracts tied to how many bots run and how many employees use them, and about 80% of revenue now comes from cloud deployments. That model depends on being the control layer across fragmented software, not inside one app.
  1 sacra 2 sacra
• The incumbents are collapsing that control layer into their own stacks. Microsoft has Copilot in Power Automate for desktop, which lets users create desktop flows from natural language, and Salesforce now lets Agentforce trigger actions inside Flow, Slack, and MuleSoft connected systems. That directly overlaps with routine RPA workflows.
  3 microsoft 4 salesforce 5 salesforce
• The market is not disappearing, it is being pulled upward. UiPath now frames the category as agentic automation, where AI agents, robots, and people work together. In practice, that means classic screen clicking remains useful for legacy apps, but the higher value shifts to orchestration, governance, document handling, and cross system execution.
  6 uipath 7 sacra 8 uipath

Going forward, standalone RPA vendors win by becoming the automation layer for the hardest enterprise environments, where workflows cross old desktop software, documents, APIs, and human approvals. The simple repetitive tasks inside modern SaaS products will keep getting absorbed by those products themselves, pushing RPA platforms toward broader agent orchestration rather than pure bot creation.