Anna Mossberg is an experienced business leader (ex-Google, ex-Telia) and a digital transformation professional, who has a strong background in driving digitalization initiatives in telecommunications and finance. As the Managing Director of Silo AI Sweden, she has seen how companies take – in her opinion – too small steps in order to become truly AI-driven. In this interview, Anna shares her thoughts and experience about onboarding AI.
Anna, as the MD for Silo AI Sweden, you want to solve the two major challenges slowing down onboarding of AI: the lack of understanding and the difficulty of finding talent. What is the one thing executives should understand about AI?
– Artificial intelligence and machine learning are technologies that will change a lot of things in a company, including processes, products, and ways of working. The change is not limited to just things, but really touches on people, their work and organizing around AI. Onboarding AI is a learning curve and it will take some time to excel in this, so it’s important to get going. The beginning will be made in small steps, but the future opportunity is substantial: For many businesses it will be about competitive advantage and survival in the end.
Many companies are currently advancing through digital transformation: Why now is the right time to start experimenting with AI?
– Companies have their data and data pipelines in a relatively good shape by now. In addition, they have access to computing power and technical knowledge that can be either internal or externally sourced. Once you start to develop AI, you’ll start to learn along the way: the development of AI is iterative by nature – the important thing is to start learning. My experience is that once you unlock the first use cases, you start seeing many more opportunities for improvements. It’s almost like you put a new pair of goggles on and view the business through an AI lens.
In addition, if you think about the timeline, you can’t close your eyes on the competition. It’s obvious that your competitors will go this way and you do not want to get left behind…
Where should I look for the first AI initiatives within my organization?
– It depends on the industry but improvements of the core product (even smaller ones to facilitate learnings and future larger steps), predictive maintenance, customer channels, recommendations, optimization and automation of processes would be my first candidates.
But in all honesty at Silo AI we have seen an immense variation of use cases by now so anything where there is a very complex task, or a very simple repetitive task to be done and where this task has a real business impact are good candidates in my eyes.
We typically start off with a design sprint to ideate different AI features with our client together. This, I believe, is the most efficient way, as we have the deep AI expertise, and our client obviously knows their business, domain and the challenges they need to solve. After that, we assess the project feasibility with machine learning through a review and create a Proof-of-Concept to validate the idea.
How is AI helping people to perform better at work?
– At Silo AI, our vision is AI for people. We strive to build AI-driven products and solutions that help people to do their jobs better and AI can help people with very complex cases where it’s not human possible to optimize or process all parameters. One example can be our work with Helen (read customer case), a major energy firm in Helsinki, Finland where the ML-driven predictions reduced the error rate between heat forecasts and actual demand by one third. Another example would be our work with the Swedish Tekniska Verken (read customer case), where we incorporated computer vision into sewage pipe inspection and predictive maintenance. In this case, AI contributed to removing simple repetitive tasks or work in very demanding environments.
In both cases we need transparent systems and explainable AI so that humans can accept the predictions and recommendations made by the AI, and understand that if the product tells the human operator that there is something in the sewage pipe, the data on which this decision was made upon can be found and be double checked by a human.
What’s the biggest challenge in becoming more AI-driven?
– To me the biggest challenge is the lack of knowledge and the hype around AI. Many people know they should do something, but they perceive this to be a very complex question and thus put it as something we should do, but later… and the train has already left the station.
Second, it’s the fact that executives might have too big expectations too soon and evaluate even early experiments as stand alone business cases that should deliver a substantial financial contribution. If you do this especially for AI development, which is an iterative process, you risk to get disappointed based on too early conclusions and a stop to the journey of becoming more AI-driven.
What would you suggest as the first steps in starting a company’s AI journey?
– My suggestion would be to identify the areas where you know that an improvement would make a difference to your business and quickly do a feasibility check for AI (Design sprint). Then scope AI use cases and decide to at least get one experiment going (PoC), the rest will follow…
Would you become more AI-driven? Get in touch with Anna at email@example.com.