The AI Aurora: How Nordic companies are taking generative AI to new places

If you look back, Scandinavian inventors have given us some of history’s greatest inventions — dynamite, insulin, loud speakers, the heart rate monitor, even LEGOs and, more recently, Spotify.

The Nordic countries are undoubtedly world leaders in innovation and reinvention, known for their open cultures, collaborative spirit, and forward thinkers. Unsurprisingly, the region has cultivated a reputation for creating next-generation services and products, driven by a deep commitment to research, development, and no-holds-barred approach to embracing the latest digital opportunities to help them achieve success.

When it comes to the next wave of opportunities, perhaps none are bigger than the latest advancements in generative AI and large language models, or LLMs. A recent McKinsey study suggested companies that successfully absorb AI tools across their organizations in the next five to seven years could double their cash flow by 2030. Generative experiences and tools will not only play a huge role in unlocking new innovation but also help make data and existing technologies more accessible to the wider business workforce.

Still, Nordic organizations have some catching up to do, as a recent Accenture report found that only 6% of companies in the region are considered AI achievers, compared to 12% for the whole of Europe. That’s not to say that Nordic countries are failing to progress in their AI transformation journeys. The same report estimated that roughly 78% of organizations have already re-worked their existing cloud strategies to better achieve AI success.

Even so, the large majority are still struggling to move beyond early pilot projects and successfully operationalize their AI strategies.

This is where generative AI holds massive potential to transform the way AI applications are built and who can interact with them – and as a region that has never shied away from emerging technologies, many Nordic innovators are already taking the lead on finding practical paths to getting started with generative AI.

At a recent Google Cloud Summit Nordics held in Stockholm, thousands of cloud experts and enthusiasts came together to explore how generative AI is reshaping both their region and the world. Here, we hear from five of them.

Happeo: Start with the problem, not the solution

Organizations that only focus on a desire to use generative AI may find it harder to move forward than those who deliberately apply it to solve a problem.

Finnish startup Happeo helps organizations build next-generation intranets that make it easy to search, find, and access information across tools and sources from a single portal. Happeo has added generative AI to its mix to provide a faster method for adding information to intranets, which helps accelerate the time and effort it takes to create and publish new content.

For example, the platform already comes with pre-built templates and a drag-and-drop builder to speed up page building, but now it also offers an AI assistant, powered by models trained and run on Google Cloud. Using a simple description of the page and the information to be included, the AI assistant can generate a suitable page layout in seconds — complete with copy, images, and metadata. It can even predict where it should be located in the intranet and appropriate tagging so customers don’t have to spend additional time figuring out where it should live.

Neurons: Enhance existing areas of expertise

Another way to help determine what use cases work best for your organizations is to consider places where generative AI can help you enhance your strongest offerings. Danish AI insights platform Neurons, for instance, is building off its existing predictive AI expertise and looking to bring even more value to its customers with generative AI models.

Neurons has already made a name for itself building powerful marketing models with Stanford University that can help predict marketing ad performance in seconds. The models, which are trained and run Google Cloud’s Vertex AI platform, can predict how well content will capture an audience’s attention, engage them, convey the message, and stand out in their memory.

Now, Neuron is building an AI solution that can generate high-performance ad recommendations based on insights from its predictive AI models. Generating ad content quickly in no way guarantees performance — especially in an industry where around 73% of ad spend already fails to make an impact. By combining generative with predictive AI, Neurons believes it can provide better personalized ad recommendations based on multiple factors, including brand guidelines, tone, industry benchmark, in-market effect data, and firmographics.

Podimo: Transform search into personalized discovery

Generative AI opens up the door to reimagine entirely novel ways to interact and personalize user experiences. In particular, enhancing search capabilities in platforms with generative AI can make it possible to search for information in a more natural and conversational way.

One example, shared by podcast subscription service Podimo, is its new conversational search experience. The Danish company has been actively working to transform search into a personalized discovery engine with semantic search capabilities and results ranked according to specific user queries and listening activity. Now, leveraging Google’s foundation model PaLM 2, Podimo has built a new conversational interface that enables users to “chat” with the search engine to find podcast recommendations.

Instead of searching for specific titles, authors, or keywords, a user can instead type in free text, such as, “I’m looking to learn something about Roman history today” or “Do you have podcasts about AI featuring top university professors?” and get recommendations that match their intent and personal preferences.

Vionlabs: Capture next-generation opportunities for engagement

Generative AI’s ability to synthesize information and generate natural language response has many applications beyond search. A gen AI-powered chatbot or assistant has the potential to simplify interacting with complex platforms or technologies down into writing simple instructions or questions — also known as prompts.

Swedish AI leader Vionlabs, for instance, has sophisticated AI solutions down to an art form. The platform, built with Vertex AI, uses computer vision and the latest AI techniques to “fingerprint” video data, extracting granular metadata from image and audio. This can help enable streaming companies and media businesses to capture the emotional impact and mood of content to boost engagement and improve personalized recommendations. AINAR, the company’s groundbreaking cognitive AI technology, can predict and produce over 40 genre types, 1500 keywords, 40 mood categories, and over 700 mood tags to observe and comprehend the media it consumes.

Now, using generative AI, Vionlabs is looking to give viewers even easier ways to find and interact with streaming platforms. With AINAR’s Prompts, users can describe the exact content they are looking for and get video content back. Instead of wasting time searching through categories or scrolling through recommended carousels, you could find relevant results just by typing in: “Give me a fast-paced action movie with lots of car chases and explosions.”

Vionlab’s innovations are showing us a glimpse into a future where many different types of users will be able to interact with and use very advanced, sophisticated technologies without having to be an expert.

Strise: Separate fact from fiction with experimentation

One common question we hear from leaders and customers is whether they can implement generative AI safely and securely. In particular, companies that operate in highly-regulated industries, such as financial services or healthcare, have deep concerns around how to use AI while adhering to regulatory requirements and data privacy laws.

When Norwegian startup Strise started evaluating how to incorporate generative AI into its anti-money laundering (AML) intelligence software, the team was also curious about the security and compliance implications. After speaking with several companies, Strise found hackathons are one of the best (and safest) places to innovate a product with LLMs. Hackathons provide developers with the freedom and space to explore and assess what works best — and what doesn’t.

The company is currently experimenting with new gen AI capabilities that can simplify the user flow, including an “Ask AI” feature that automatically generates triggers for monitoring companies. For instance, a customer could type in the goal of a trigger “three months since last review and high risk score,” and Strise will automatically populate the correct condition options to trigger an action.

By adopting a regular practice of internal experimentation and brainstorming, regulated companies can still drive new innovation and discover new ways to help their customers solve critical problems with gen AI without increasing their risk of exposure or regulatory violations.