Turning AI into Value: From Data to Impact

That was the central question when Epinova welcomed clients, partners, and peers to a breakfast seminar at NOVA House Oslo. The session explored how generative AI is already reshaping the workday - from organisational models and data foundations to search and visibility.

The goal? To understand how businesses are creating real value with AI - not through trends, but through tangible actions, frameworks, and cross-disciplinary learning.

How Avinor is operationalising generative AI

When you manage 51 million passengers annually, 43 airports, and 3,000 employees, AI isn’t something you simply “try out.” Petter Norstrøm, Head of Internal Communications and Digital Channels at Avinor, shared their journey to embed AI into everyday operations.

It began with curiosity and evolved into structure. Avinor quickly established an internal AI group composed of engaged employees from across disciplines. Empowered to experiment, test tools, and share learnings, the group helped spread knowledge safely throughout the organisation. The result is a culture of rapid learning, open sharing, and continuous improvement.

“We have 742 licenses for nearly 3,000 employees and 90% are active users.”

The key, Norstrøm explained, is combining curiosity with accountability. Those granted access must demonstrate real usage and understanding of the tools. At the same time, the threshold for experimentation is kept low – encouraging people to try, fail, and share in safe spaces.

In the communications department, AI is already supporting tasks like text refinement, translation, and presentation development. The larger initiatives: data-driven traffic planning, optimal gate allocation, insight-led winter operations, and automated invoice control demonstrate how AI creates real operational impact across Avinor.

Most importantly, Avinor has succeeded in making AI a shared language across the organisation – not a separate project, but a natural part of daily work.

Key takeaway: Build AI from the ground up. Enable experimentation, but ensure clear frameworks and ownership.

Data First—Always

“Without accessible, high-quality data, AI simply doesn’t work,” said Kristian Borg, CTO at Epinova.

While Avinor demonstrated how to organise AI, Borg took the audience deeper – into the prerequisites that must be in place for the technology to deliver.

He described a landscape where many organisations sit on vast amounts of information – yet much of it is unstructured, siloed, and lacks clear ownership. The result? Generative AI becomes as fragmented as the sources it learns from.

Success starts with a “data-first” mindset: structure, accessibility, and quality assurance must precede the implementation of new tools. This is not just about system architecture – it’s about culture and process.

Borg also looked ahead to API design for the future – how organisations must enable communication between APIs so that data, information, and services can flow seamlessly, including to and from AI systems. As AI becomes more integrated into how we search, ask, and receive answers, APIs will form the foundation of data flow.

He closed with a thought-provoking question:

“What if our websites eventually become secondary interfaces—while real communication happens between AIs?”

Key takeaway: Build a robust data foundation and design for a future where APIs communicate as much with AIs as with humans.

Navigating AI-driven visibility

Next, Erik Egeriis, Head of SEO at Rocket, addressed a pressing question: How do we succeed with visibility in the age of AI Overviews?

Drawing on Rockets comprehensive study of over 100 million searches and data from 130 Norwegian websites – before and after the launch of AI Overview – Egeriis showed that the reality is far less dramatic than headlines suggest.

Yes, traffic patterns are shifting – but not in ways that threaten website relevance. In fact, they open new opportunities.

Rather than fearing AI-generated search results, businesses should view them as a new arena for visibility. To succeed, content must be clear, specific, and structured so both humans and machines can understand it. That means using explicit language, accurate metadata, and structured data (schema) to help AI interpret context.

“See it as a space of opportunity – not just a threat.”

For marketers and editors, the message is clear: instead of writing for search engines, we must start writing with them.

Key takeaway: AI doesn’t diminish the need for great content – it amplifies it.

From experimentation to impact

The seminar brought together three distinct perspectives on a shared challenge – offering a holistic view of how organisations can meaningfully adopt generative AI.

The common thread? Value is only created when people, processes, and technology work in harmony.

By bringing together clients, partners, and experts at NOVA House, we create space to share experiences that help move from experimentation to real impact.

At NOVA Consulting Group, it’s not just about testing new technology – it’s about building it on the right foundation.

Data first. Always.

The seminar underscored why data discipline is the key to AI success. When structure, architecture, and insight are in place, technology can truly create value.

That’s where we want to be – alongside organisations ready to use data as a driver for real impact.