Coffee break with Elly Geck

Engineering the Future: Elly’s journey from processes to AI

Interview: Ceren Gökhan

Interviewee: Elly Geck

This edition of Coffee break features Elly Geck, our Internal Automation Team Lead. Having been with Solaris for over four years, Elly has seen the company grow through its many ups and downs. She shares her perspective on why solving problems is the most rewarding part of her job and how she navigates the fear surrounding AI.

Elly, you’ve been with us for over four years and have seen the company grow and change. Looking back, what has been the most rewarding part of moving from fixing processes to now leading a team that builds the future of how we work?

Elly: For me, what I always focus on is working with nice people and being able to solve problems. This is the most rewarding part: when I see that I am solving an actual problem for someone. I think that be it in product management or also in engineering, you are in a really comfortable situation because everyone is really happy when you go to them and ask them how you can help.

I was doing this partly in my product role where I worked a lot on improving processes, and now I have moved to this role of process engineering where the whole job is actually doing this. We go to people, check their problems, and try to understand what is the most high value problem to solve. I am really lucky working with such a mature team of senior engineers. We now have the bonus of doing this with AI, which is like a whole new toolbox that has opened up. You can really see the effect of the work you are doing because you are making things easier for everyone.

We have so many different departments and tasks across the bank. As a leader, how do you and your team decide which manual process is the next one you’re going to solve with automation?

Elly: This is a classic product management task. Prioritization is key because if we do not work on the right things, we are not going to move in the right direction. We have a lot of people coming to us with small issues, so we always need to ask what the business case is.

We look at how much time we are saving and what the benefit of solving that high value problem is. We also look at how complex or urgent it is. Sometimes it is not even about saving time, but about limiting risk. We have moved quite a bit toward this risk based thinking, so limiting risk is another really good use case for us. We do a value urgency exercise and based on that we decide which ones to take on. Honestly, sometimes it is really hard to say no, but we need to say no quite a bit.

There’s a lot of noise about AI these days, and it can be a bit overwhelming. You’ve said before that your goal is to help, not replacement. How do you explain to a colleague that these new tools are meant to be their superpower rather than something to worry about?

Alicja: First of all, we need to be okay with the fact that people are scared. Even I sometimes get scared. Software developers get scared. I think everyone probably gets scared, and that is totally normal.

I always try to say that there is only one way and the way is forward. What we can do is give people the opportunity to dip their toes into AI and really understand what they can do and what the limitations are.

We just need to go slow and co-create with the teams. For example, one of our huge cost savers is ticket triage in customer support, making sure a ticket goes directly to the right category. We are in a regulated environment, so for us, we will definitely be the ones with the human in the loop. There is always a human who needs to take a decision. It is augmentation, not replacement. We are helping people to do their jobs faster, not replacing them.

What’s the most frequent challenge your team encounters, and how do you typically navigate or solve it?

Elly: One of the main challenges is simply the variety of ways we can automate things. We have a lot of different options, from connecting APIs and building super custom software in engineering to using tools like Zendesk. For example, we have a Zendesk administrator on our team who can configure integrations like Trustpilot without needing to program, but you still need someone who knows the software really well to do it.

A frequent challenge is choosing the right approach for the groundwork. We used to have software in operations like a UI Path, where people could automate processes on UI, but we decommissioned that. Now, we focus on automations that cover use cases end to end implementing from scratch and connecting tools as needed. Navigating this means understanding when to build something from scratch and when to use existing tools to get the best result for the company.

In your world, what is one seemingly small digital detail that you believe has a surprisingly big impact on how an employee feels about their efficiency at work?

Elly: Bringing Google Gemini to the company has been a long story. It took quite some time to get it through because we needed an AI policy, a procedure, and we had to push it through data protection. It might seem like just one more software tool, but it has a huge impact because it changes the whole way people work.

For a lot of departments, especially product, writing documentation is a huge part of the job. Now, you have someone to brainstorm with. People come to me and tell me about great use cases where they use AI to handle internal information that they wouldn't have been able to use with other external tools. Seeing how much happier people are because they have this assistant to help with reporting or benchmarks is really great. It is a digital detail that has turned into a real superpower for our employees.

When tackling a pain point process, what’s the most frequent challenge you face, and how do you help teams feel comfortable transitioning to new AI-driven ways of working?

Elly: The most frequent challenge is actually understanding what the real issue is. It is often very hard to pinpoint the true pain point and even harder to define how much value a solution will bring. Normally you do not have exact data, so you are just expecting that a process will be faster without knowing for sure.

To help teams feel comfortable, we go slow and focus on co-creation. We work with them together so they do not feel like a new way of working is being pushed on them. One thing I worked on was helping colleagues who had to check documents, which is very monotonous work. It took some time until they were fully convinced, but by letting them pilot the technology and giving them access to tools like the AI agent platform I am testing now, they can start to have their own ideas. When they see the benefits for themselves, like how an agent can check legal tickets or regulatory changes automatically, the transition becomes much more natural.