What distinguishes Agentic Automation from RPA?

RPA (Robotic Process Automation) executes strictly defined rules and click sequences: the robot does exactly what was modeled beforehand and breaks as soon as reality deviates from it. Agentic Automation instead sets a goal and leaves the path to the agent – it selects tools, reacts to unexpected states, and re-plans. The difference is therefore not the degree of automation but whether the sequence is prescribed or decided at runtime.

Why does this matter?

RPA typically fails on exceptions – and in most processes that is exactly where the greatest effort sits. Agentic approaches cover this exception space, but demand new control mechanisms in return: whoever does not prescribe the path must be able to review it afterwards.

What does this look like in practice?

A supplier changes the layout of its invoicing portal. The RPA bot clicks into empty space and stops until someone updates the script. An agent reads the page, recognises the moved fields and carries on — but reports the change, because it is a deviation from the known flow.

Not to be confused with

Not to be read as a replacement: RPA remains the better choice where a flow is stable, high-volume and precisely specified. Agentic automation pays off where variance and judgement are involved — in practice the two often run side by side.