Riku Ruokolahti: "AI gets more out of data when the framework is right—we built our own secure environment for this purpose"

Reputation and Trust Analytics has developed its own AI solution that leverages open-source models to interpret data directly through Reputation&Trust. When the AI operates within our own closed infrastructure, the data remains secure and the analyses are precisely contextualized. Riku Ruokolahti, Head of Development, explains why AI is at its best when clear workflows have been established for it.

Riku, how and why did Reputation and Trust Analytics start developing its own AI?

In 2023, we had a vision that we could use artificial intelligence to synthesize the quantitative data generated by Reputation&Trust with the open feedback—that is, the qualitative data—collected through our research.

We wanted to leverage the opportunities offered by artificial intelligence both to improve our own operations and to provide our customers with even greater benefits Reputation&Trust.

Right at the start of the project, we explored whether we could use any third-party services, but it quickly became clear that this wasn’t really a viable option. We would have had to hand over both Reputation&Trust and our customers’ data to the AI. That is, of course, out of the question. We will not hand over our most important intangible asset—our customers’ confidential information—to third parties any more than we would our own.

So we came to the conclusion that we must have artificial intelligence under our own control and the ability to use it under our own roof.

How did the project get started?

We began working to improve the quality of our qualitative data right away in 2023. We redesigned our data collection methods and questions to ensure a richer context for when we eventually get around to putting our dreams into practice.

Next, we began experimenting with various AI models. However, they are evolving rapidly, as is the hardware, which accelerated the development of our own AI solution—but at the same time also posed challenges. We couldn’t really commit fully to any single AI model, since a new and better one might already be just around the corner.

To tackle this challenge, we created a process that we can continuously refine. Thanks to this, we gain access to new, better, more concise, and more effective templates as they become available.

Quite a few of our employees have been involved in the AI development project at various stages, but our senior development planner, Jaakko Reinikainen, has done a great deal of the hands-on work.

Where are we headed—what does our own AI look like right now?

I’d say we’re now in a pretty good place with AI. We’ve built the infrastructure, operating models, and processes needed to run local AI models. Our original vision, in which AI combines quantitative and qualitative data, is now working very well. AI efficiently analyzes open-ended feedback through Reputation&Trust and generates actionable recommendations based on existing data, supported by quantitative data analysis.

Naturally, the results must be refined, taking into account how they might align with each company’s strategy. AI will not take over the management of the company or its reputation; rather, this intellectual work remains the responsibility of the company’s leadership. However, the results generated by AI serve as a solid foundation for the thought process and provide management with interesting insights that could not have been discovered through thought alone.

Artificial intelligence and human thinking each have their own roles to play here, where they are at their best.

What happens next?

Of course, we’re always keeping an eye on what’s happening in the world and in the field of generative AI.

We continue to follow the two paths we set out on right from the start. The first path is our commitment to developing our own operations using the opportunities offered by artificial intelligence. Second, we are constantly thinking about what new and valuable things we can deliver to our customers with the help of our artificial intelligence.

Artificial intelligence is a particularly valuable tool for us, as we have access to vast amounts of comparable data that the AI model underlying our research understands very well. We conduct our research with extreme process discipline and precision from the very start of data collection, so we are in an excellent position to leverage artificial intelligence. We know and understand what our artificial intelligence does and how it benefits us.

Riku was interviewed by Anu Räsänen, a communications specialist at Reputation and Trust Analytics.