Hey! Pietari Pikkuaho here, Head of AI at Capful. I have been working at Capful for two and a half years, of which the past two years has been on building our GenAI practice from zero. I wanted to share some reflections on working in the fastest-moving field I have ever experienced, and highlight a few things that have stayed relatively constant despite the pace of progress.
The weird thing about working with AI is that even if you believe in the rapid development of AI, you might still end up being surprised when that pace materializes, because you have failed to internalize the belief. While a lot of our strategy in the AI space has been founded on the belief that GenAI will continue to scale and that we will see rapid development in the coming years, it is one thing to say you believe that. It is another to wake up every six months to a paradigm-setting breakthrough, that requires you to change direction or pivot, because the problems you have been solving have been either solved or invalidated by the technology.
It is also what makes this by far the most dynamic job I have ever had: the field keeps changing, and so must your strategy, roadmap, and playbook.
Still, there are a few things that have not (yet) been invalidated about working with GenAI. One is the importance of benchmarks, evaluation, and validation. You need to be able to concretely measure the performance of a new model or tool at a task and benchmark it against what you had before. Moving to a new model is rarely a simple drop-in operation. Different models still have their own quirks and often require extensive validation to make sure they do not break your existing tools.
Another is that codifying institutional knowledge is real, and your AI solution is fundamentally bounded by your own capabilities. If your team can deliver foresight at a level of 7 out of 10, your AI tools will not magically reach an 8 – because you do not know what an 8 looks like, and neither will your AI tool. This is also why traditional subject matter experts are not just helpful in building AI solutions, they are essential. The best AI tools emerge from marrying domain experts who understand the craft with technologists who understand the technology, and involving both in designing the solution.
For us, the past year and a half has been a labour of building toward what we see as the future of foresight, and strategy consulting at large. We launched CapfulAI, our internal GenAI-driven foresight platform, in the spring of 2025. Alongside this, we completed a 9-month Business Finland-funded research project exploring agentic reasoning, and are now participating in a new research initiative led by researchers from Aalto University and Oxford examining how AI is reshaping consulting. Looking back, I am incredibly happy that we started as early as we did, because we have needed this time to construct a solid foundation and lay the groundwork for what comes next. This also allows us to move from the foundational phase to the transformation phase, where we can truly start to see what the next era of consulting might look like.
We see the future moving in two directions: a platform-based future where consulting companies codify their knowledge into digital platforms, and an AI-driven hybrid future where small, nimble teams leverage AI to punch above their weight and solve complex problems.
Our aim is to work in both. Through CapfulAI, by codifying over 25 years of institutional knowledge into a platform-based offering, and by leveraging the tools we build to deliver best-in-class consulting and solve the most complex problems in strategic foresight.
After a period of quiet building, we are ready for what comes next. The next six months will be some of the most critical – and exciting – in Capful’s history, and we cannot wait to share what we have learned on our way here.
Pietari Pikkuaho
Head of AI, Senior Advisor
+358 40 124 1471
pietari.pikkuaho@capful.fi
