The time of AI pilots is over: Gabriel von Mitschke-Collande on scaling AI (Video)

“From my perspective the time of pilots is over,” says Gabriel von Mitschke-Collande.

He is not suggesting organisations stop experimenting. His point is that once AI has proved it can do something useful, a different set of problems appears. Licences and computing cost money and organisations need to know what they are trying to change and where the value will come from.

Giesecke+Devrient has been working with AI since 2017 and 2018. Its early activity was concentrated in the tech office where specialists explored use cases and learned what the technology could and could not do. As interest accelerated after ChatGPT arrived, G+D decided that approach would no longer be enough. Technical knowledge needed to be combined much more closely with knowledge of the business.

For employees von Mitschke-Collande says this means being honest about what AI may change and giving people the chance to gain their own experience of the technology. G+D provides tools employees can try in their work alongside training and learning designed to build AI literacy.

But the more difficult work is beneath this – what von Mitschke-Collande describes as the AI “homework”. Organisations need usable data and systems that can provide access to it. Processes need to be understood before they can be automated and some workflows will need to be redesigned rather than simply handed over to an AI agent.

This has consequences for work itself. As workflows change von Mitschke-Collande expects roles to change with them, which makes leadership and people capability an important part of using AI more widely.

For HR and business leaders still working out how to move beyond isolated pilots G+D provides a useful example of what comes next.

Key takeaways from this In Practice video

• Successful AI pilots are only the beginning. Organisations still need to decide where AI will create enough value to justify further investment.
• Employees need access to appropriate tools so they can experiment and understand where AI helps and where it falls short.
• AI literacy requires more than general awareness. People need to understand the tools their organisation has chosen and how they can use them safely.
• Data, IT systems and clearly understood processes become increasingly important as organisations try to automate more complex work.
• Workflow redesign and changing roles make AI adoption a leadership and people issue as well as a technology one.

Watch the full video above to hear Gabriel von Mitschke-Collande explain what G+D has learned as it moves from AI experimentation towards using the technology at scale.

Read the full story: Scaling AI at work: what G+D learned beyond the pilot stage

Transcript: Gabriel von Mitschke‑Collande - Group CDO of Giesecke+Devrient

From AI pilots to scaling AI across the business

From my perspective the time of pilots is over. And yes, of course, we need pilots and we need to come up with new ideas, but we need to generate value because, on the other hand, AI is does also not come for free. So licences cost. The discussion of the time is token consumption. So all this cost extra in our P&L. So yes, we must talk about scaling, and we must talk about value creation that we in the end have a positive ROI as an organisation.

So we started AI, our first AI activities, back in 2017, 2018. So fairly early. So prior to this ChatGPT hype, when it really became a public-known technology. And we started it in our tech office.

So a bunch of high-technological, high-educated colleagues who looked into this technology and came up with first use cases. And this was super important to learn what it requires to use this technology, also how it works, what are limits, what are potentials and so on. And latest since 2022, when ChatGPT came up, we figured out that we need to take different steps forward in order to scale because it was not a question of what individual use cases we need to look at. It's really now the question, how can we scale the usage of this technology and how can we generate value with AI? When I talk value there are three pillars to think about. One is how can we increase productivity? How can we generate new revenue

in the combination of current product portfolio and AI, and ultimately how can we build additional new business, AI-as-a-product approach. And here is something where we needed to put the technology perspective out of the tech office and combine it with a business perspective. And this is where we came up with we need to change our structure, how we do AI.

Why AI literacy and workflow redesign matter for adoption

So here, I think, well, the levers, isn't it? So one is the communication part. And we cannot over communicate AI. And yes, there are also some fears. If you read the current headlines in the newspapers, yes, there are also some questions of fears. So what does this mean for us as a company in transformation always creates questions, positive ones and maybe some critical ones. And we as management, we need to answer them in an honest way. So this is number one. 

Then number two I think accessibility to tools is important. So we must provide tools that our colleagues can work with it and also can play around with it. So everyone needs to experience what is the potential of it? Where are also the limits? How can I gain my own experience in terms of, okay, just, you know, training on the job? I have a tool, I give it a try, I learn something, I get the result, the results help me or I need to rework the result. So really provide an environment where people can give it a try. 

And thirdly, I think training is super important. So working on the AI literacy is important. So what training offerings do we provide, which is for sure different than Project Management 2, which is a different type of a course than working with AI. And here we have collaboration with the institution which is close to the Technical University Munich. We have our own developed training modules, which we provide via our intranet. And all of the above is important to really provide an offering that people can learn, can try, and then ultimately can also learn to understand how it works and what is the potential we are looking at here.

Access to talents is crucial. It is a new discipline. There are talents within our organisation, but also outside our organisation. And we must make sure that we have these talents on board because we need the talents in order to come up with high quality solutions in an adequate pace and speed. So this is number one. 

Number two is often we talk about the magic which is doable with AI. So we all know the headlines that we increase speed and that we increase and improve quality and all these things. But what is also true is that we need to talk about the homework. So how are our data structured? Do we have the data? Do we have access to the data required? How does my IT architecture look like? And if I have different systems where I need the data from, do I have access to the systems? So all of this foundational work also needs to happen in order to build scaled solutions on that. Part one, part two, we need to talk about processes. 

Today we talk about AI agents and that AI agents can basically automate processes. So if you want to automate processes we need to have documented and defined processes. And if we don't want to automate process step three to four, then we also need to redesign our workflows. So this entire process engineering discipline becomes more and more important again, because we need to redefine the way we work together. So this is also something we need to think of, and we need also to invest into because without this investment we don't end up scaling here. 

And yes, leadership is important. We must have a good leadership around this topic, a) for building a vision, this is part one, but then also translating the vision in concrete measures and also in communicating where we are on the journey and what we have achieved and what is still missing and also listening to the organisation where are some concerns or where does our message not reach the organisation or where are great ideas which need to flow up the organisation because this is a huge opportunity for extra business. And this is a leadership topic which we also need to consider and need to foster because AI in the end is a team play and we need every hand on deck in order to make it successful.

Three practical steps to scale AI responsibly

So my first recommendation in terms of actions is make it a priority. So this technology is a priority, then you can actively work on your own transformation. If it's not your priority you won't get the value out of it. Number one priority. 

Number two is really to consider the question what do you want to change and what do you want to solve? We must have a clear target picture in our minds for a use case and also b) for organisations which we want to transform based on that technology. 

And last but not least is, of course, resource allocation. And this is linked to number one, the priority. If it's a priority we must allocate resources to it in terms of expertise, in terms of talents, but also in terms of budget, because we need to actively invest into it. And I think these three topics are super important. And the foundation here, of course, is keep the colleagues in the workforce in mind because we need the organisations to make the transformation happen.