AI in communications: three steps from experimentation to competitive advantage

Ten years ago, companies were already using AI in many areas, from predicting customer churn to industrial automation. Its role was narrow, but the growing significance was already anticipated.

“Data and the AI that leverages it will undoubtedly become one of the key competitive differentiators between companies,” said Harri Nummela, Head of Digital Business and Customer Experience at OP Financial Group, in my blog back in 2017. In 2026, OP is one of Finland’s frontrunners in AI, with an ICT development budget of nearly half a billion euros.

In communications, however, only the first steps were being taken – for instance, YLE was at the forefront of content automation. Few tools were available, and most had to be coded from scratch. ChatGPT would not launch for another five years.

At the time, I wrote that “a conceptual understanding of possibilities matters more than technical expertise. Technology can already be bought in bulk – understanding what it can do for you cannot. If you know what you want to achieve, finding or commissioning the right application is the easy part.”

Ten years on, the range of tools has exploded, but that observation could not be more true. No one in 2017 predicted the wave of vibe coding or the rapid emergence of agents that handle office workers’ tasks – yet the need to understand your own and your team’s work processes remains the starting point for everything. 

Tier 1: Ad hoc assistance 

A language model or similar tool helps an individual expert with specific tasks, such as summarising large volumes of material or drafting a LinkedIn post.

Tier 2: Partial automation 

AI, scheduled tasks, and agents or “skills” shared across the team do the groundwork. For example, automated monitoring of relevant material and stakeholder analysis deliver ready-made summaries to the communicator’s desk every morning – even in the form of a PowerPoint presentation.

Tier 3: Low-code pipelines 

Tools like n8n, Zapier, or Make.com automate entire processes end to end. For instance, a draft position paper or blog post is produced through automated monitoring of source material, integration of internal and external databases, and stakeholder modelling. After human editing and approval, the automation publishes it across selected channels in the appropriate format for each. The automation also handles impact measurement and reporting.

In 2017, this was a dream. In 2026, all of it is already happening in dozens of Finnish companies. The key point is that no one needs to jump straight to tier three. Integrating AI is a cultural shift, not a technology project: it requires new capabilities, a willingness to challenge established ways of working, and the courage to let go of familiar routines. And above all, hands-on use of the tools yourself.

That is exactly why a step-by-step approach works. When a team first learns to use AI on an ad hoc basis, the threshold for the next tier drops – and at the same time, a shared understanding emerges of what is worth automating and what is not.