For the AI Strategy Lab project, I took a less leading role than I usually take in group projects, simply because from my initial interactions with my group mates, I realised that (to my relief) they were going to carry their own weight and not sit idle waiting for others to take action or choose for the group. Regarding the project itself, in the beginning and around the middle, I was confused about what we needed to accomplish and how exactly we could use what we learned during the lectures. One of my main worries was the ‘prototype’ we supposedly needed to make. I cannot express enough how little I know about coding or creating and managing software, and for the first couple of weeks, I believed we needed to build a fully functioning product for the final project. Thankfully, I chatted with fellow students and realised my worrying was for naught.
I had a few ideas for our project, but given my inexperience in the working world and actual businesses, they were not chosen. Instead, we went with our team mate’s idea to help those in the medical field. However, I did not have much relevant experience in that area either, so I could only research a few topics and possible ideas we could move forward with. The team mate who proposed the healthcare idea came up with a really good project concept, which interested me because she had actually asked doctors and therapists at a rehabilitation centre about their problems, what they wished could be improved, and what would make their work easier. We decided to focus on workflow optimization, which could substantially reduce administrative work and paperwork and relieve some of the workload on staff.
At the start of the project, before our idea was completely defined, we were inspired by Augmented Reality devices discussed during the lectures and considered how we could use AR to improve workflows for warehouse workers and distribution centres. One of our team mates worked in logistics and was excited about her idea, believing it could be extremely helpful for workers. However, the next day, she found many examples of this idea already being implemented in several companies, so we moved on from it. The third team mate, like me, did not have many concrete ideas she wanted to explore, so the group chose to focus on healthcare workflows, and we gave it a very original name: Careflow.
When we started, the idea was not well-defined, and we had not decided exactly which processes we wanted to improve. The problem was that healthcare professionals face administrative burdens and inefficient workflows that consume time and resources. The market opportunity we saw was that AI adoption in healthcare creates an opening for solutions that can be implemented relatively quickly to address specific workflow problems. Our proposed solution was to use Agentic AI to automate administrative workflows that add to the already heavy workload doctors face. However, we knew implementation could be challenging, given that healthcare and patient medical data must be protected, kept secure, and only accessible to authorized people. The value it could bring was allowing doctors to focus more on patient care.
Working with my team mates went mostly smoothly, with the exception of the last couple of days before the deadline. We did not receive feedback from the lecturer beyond the first session, which helped us focus our idea because, as mentioned before, it was still vague and needed more refinement. Most of our feedback came from friends and classmates whom we asked about their projects and what we should focus on, since we were a bit lost.
Around the middle of the project, we realised we were relying more on speculation about administrative workflows and how they could be optimized than on the exact problems and time-consuming tasks we wanted to address. Thankfully, our team mates had direct contact with healthcare financial managers, doctors and administrative staff. This allowed us to focus on prescriptions for patients’ treatment plans in rehabilitation centres and how we could optimize that process.
My contribution was less than I originally planned or wanted it to be when it came to building the prototype. Since I have basically no experience creating software products or applications, my input was limited to simple user testing, such as checking whether buttons could be clicked, whether anything changed, and whether it worked in general. My focus was on the business ecosystem and how Careflow would operate within the constraints of the healthcare field. We concluded that Careflow should be considered an orchestrator platform because it manages data flows, and I illustrated this with a diagram. I also looked into how user demand triggered the AI and how users received and created value through Careflow. Additionally, I researched the suppliers and integrations needed to deliver the solution while complying with Dutch Healthcare Authority regulations, privacy requirements, and expectations of responsible AI use in healthcare administration, which falls under the high-risk classification.
I also looked into Careflow’s network effects and data flywheel, and how more users could improve the platform through feedback, workflow optimization and better integrations. However, I identified limitations, especially regarding healthcare data privacy and the possibility of insurers adjusting regulations if too many rehabilitation centres started optimizing their DBCs. The part I enjoyed most was creating and organizing the slide deck, making it visually pleasing while still conveying enough information for readers to understand Careflow without needing extra material.
My understanding of AI before and during the initial lectures was that it was mainly a tool for completing a few tasks more easily. After the lectures and towards the end of the project, I better understood how AI can optimize work and reduce workloads. I learned that it should be used as a tool to give people more time, rather than as a substitute for them. This became my main conclusion about the value AI can create for people and companies.
In the future, I will definitely start working on projects in small increments from the beginning of the block instead of leaving work until later and completing it in long, continuous batches. I also want to communicate more proactively with group mates, even though communication was not bad this time. Lastly, one perspective I gained was that technology regulations can significantly shape how technology is applied and used. It is something I need to research from the beginning; otherwise, I may end up having to rewrite much of the work.
