The world of automation is undergoing a quiet revolution — not by replacement, but by fusion. Once neatly separated into categories like Test Automation, RPA, Application Integration, and more recently, LLM & Agentic Workflows, these domains are beginning to overlap in unexpected and powerful ways. Yesterday I attended a session of our Test & Quality Guild. A pre sales engineer from UIPath was present — and I did not really understand why. Then he started to present: Test Cloud from UIPath — a Test Automation tool leveraging the same automation and scheduling capabilities known from the RPA platform I have UIPath always known for — and now empowered with agentic capabilities for creating the automations, running them more flexibly and autonomously and processing & analyzing the results more thoroughly.
Earlier I saw a demonstration of n8n embued with LLM Language processing and Agentic taks handling capabilities for automating a business workflow. Fields that once seemed quite unrelated are now getting closer. Experience, good practices, tools from one area can quickly come relevant in another. Technology is breaking down the silos. Now we humans have to follow.
In the image above I have tried to visualize what I see in my mind’s eye. Tools traditionally rooted in specific domains are now bleeding into others. UIPath, once a pure example of Robotic Process Automation, is now making moves into Test Automation with its Test Cloud. Meanwhile, Power Automate, originally an RPA and workflow tool, is integrating LLM capabilities via Copilot Studio — bringing AI into day-to-day process automation meanwhile making use of integration connectors we also know from Logic Apps.
Simultaneously, integration platforms like Make, n8n, and Zapier are evolving from simple triggers to supporting multi-step, conditional logic, and now even AI agents that can dynamically reason and act. On the frontier, frameworks like LangChain, AutoGen, and CrewAI are redefining what workflows can do — introducing agentic reasoning, memory, and dialogue to the heart of automation logic.
User interfaces are now also programmatic interfaces — because browser automation tools and agents and automatically use them to retrieve and manipulate data. The MCP “standard” (all of five months old) exposes actions, APIs, automated User Interfaces, LLM capabilities as tools for use of agents — as well as automated workflows. The Google initiative A2A will help to build cross enterprise automations | workflows | integrations …
The convergence is clear: APIs, UIs, agents, and humans are all becoming nodes in the same dynamic automation fabric. Whether you’re testing, orchestrating apps, automating manual tasks, or building AI-driven agents, you’re working toward the same goal — intelligent, adaptive, and collaborative systems that performs tasks we once had to do a lot of manual typing for.
In this emerging landscape, the boundaries are less important than the possibilities. The future is not about which tool you choose, but how well you combine them. We as developers, analysts, architects, IT specialists have to understand the potential available to us. Find a way of designing outside the box we have been in for many years — because in reality it no longer exists. Find new, cross area good practices. And do it in a responsible way: think about security and privacy, energy, accountability, social norms and regulations. Make sure the proper guard rails are in place — to reign in both the agent forces and ourselves.
