Process automation: where to start without stopping operations

The problem isn't adopting automation, it's scaling it

McKinsey's The State of AI 2025 survey shows that 88% of organizations already use AI in at least one business function, and 72% already use generative AI — up from 33% in 2024. Adoption has stopped being the bottleneck. The bottleneck is what happens after the pilot: the same research shows most companies still haven't begun scaling automation across the operation, and only a small fraction can point to real bottom-line impact from it.

That gap between "using AI for something" and "automation that sustains operations" almost always starts with the choice of the first process — not the technology picked afterward.

How to choose the right first process

The ideal process to start with has three traits: high repetition volume, a clear rule (it doesn't depend on case-by-case judgment), and an owner who feels the pain daily. Finance and back-office processes tend to score well on all three at once — high volume, defined rule, and an error that's costly and visible when it happens manually.

Processes that depend on constant exceptions, complex human judgment, or that don't yet have a clear operational owner tend to be the worst starting point, even when they look impactful on paper.

Why so much automation doesn't sustain the promised gain

A meaningful share of companies investing in automation see no noticeable cost change after the project. The most common causes aren't technical: a poorly chosen process, change management that never happened, and integration complexity with a legacy system nobody mapped before automating.

Automating a badly designed process just makes the error happen faster. Cleaning up the process needs to come before automation, not after.

What changes when the first process works

A successful first automation case works as internal proof of concept — it shows the rest of the company what "automation that actually works" looks like in practice, and builds the business case for the second process. That's why the choice of the first process matters more than the choice of the tool.

Where Diglion comes in

Diglion helps map and prioritize processes by real impact before any automation, so the first pilot becomes a base for scale instead of another project that never leaves the slide deck.

Sources consulted

• McKinsey, The State of AI: Global Survey 2025, retrieved 2026-08-11.