First Chapter
When I first started BIM and especially this AI Strategy Lab Project, I thought about very huge, superb and complex systems; for example, a rocket, robots people use for space exploration, or AI that performs surgeries on human. I did not have many expectations about an information strategy course. I wouldn’t like it that much, though – that was my first thought. As a business student from RSM and a not very technical person, I have just a few experiences with using AI, mostly in support of assignments and exam prep. Systems and AI seemed boring to me, coding was threatening to my life, and ChatGPT, Perplexity, and Copilot were no more than helping me pass exams.
Second Chapter
It was hard in the beginning to choose a problem. We went with the basics, from one of my teammates’ experiences with flight delays. My teammate encountered a long delay on her flight to the US, which caused a lot of frustration for passengers. I also shared a similar experience as a frequent air passenger – an international student from Vietnam. On my flight to AMS this summer right before the Master, my plane took off late, causing it to land later in Dubai – which made the transit for the people who would fly to AMS tight, and the airline experienced some disruptions in handling passengers as well. I found it great that we had a problem that sat near our heart and soul.
Working on AI
We found a pain point in the airline’s existing system: the processes are delivered mostly manually and inefficiently. When many passengers require assistance simultaneously, employees must gather information, search alternatives and coordinate hotel, compensation services person by person. We found that Agentic AI can greatly help automate these tasks. With great help from the technical teammates, we successfully made the prototype, with dashboard for airline staff and app for passengers. Despite having an innovative solution, the challenges were also phenomenal. Our system does not have network effects, as it is not a platform with multiple users, but an AI system for one organization – the airline. I personally enjoy session 5: Digital Platforms the most. To me, platform effects and optimizing this to make profit for a company, or bring good to people, is a promising idea that interests me a lot. Hence, I thought it would be nice if our solution had network effects. Perhaps, like a learning platform for students or an online trading platform for clothes or music. I had this interest way before starting the project, even from my bachelor’s. But the idea about Agentic AI to solve flight disruptions was pitched during the development of AI Strategy Lab, and somehow it came to us as it was. I also have a deep connection with the idea as a frequent air traveler. So the learning experience was very fun as well. I can imagine one day this product is present at airports, on a flight to Vietnam – if ever a delay happens, my alternative flight is already displayed in the app on my phone within minutes, with coupons for food and hotels, 50% off on my next booking as a recovery gift, and 100% refund of the delayed flight…all powered by AI.
Collaboration and Demo Day
I worked on the Economics of AI and recommendations for future scaling. While working together, we all had different perspectives on how to write these parts, how to design this solution, etc., as our backgrounds and studies were different, and each of us had unique way to approach problems. Pricing the AI, and reasoning the costs was the hardest part for me. I myself found it hard to convince my team that: It was not viable to find a source or academic article to reason exactly the initial development cost of 700k EUR, as this is a new AI that will be launched, not one that sustains in an existing system. From a business perspective, we should be as open and creative as academically critical. I think having solid accredited article to back up each argument is not relevant for a part like cost of AI – where we are the entrepreneurs/ creators who price our product to test and scale. I would base it on business knowledge. So in the end, I reasoned it in my own words with a small source: based on enterprise AI implementation benchmarks ranging from $600k to $1M for large integrated AI systems, we estimate an initial development investment of €700,000. A €700k development cost for an agentic AI system sits near the lower-middle range of enterprise AI costs and is realistic for a first deployment within a medium sized airline.
Given the fact that this is an AI information strategy course and the project is experimental AI Strategy Lab, I would recommend viewing the learning opportunity as trial and error. No one can give exact answers to every problem. As entrepreneurs, we should not be afraid to make wrong estimates and assumptions. Over the course, my team tended to agree with each other’s ideas, so sometimes, we neglected reflecting on our own points. I would expect my team to be more courageous in challenging each other’s viewpoints and to give compliments as well as difficult feedback more. Overall, we all worked hard on the deck – I had fun working with my team on our solution, and it was a great chance to learn about each other. I honestly feel that through this project, AI brings us humans closer to each other rather than separating us (as said by many). I had never talked this much with this many people one year back in my gap year…
Last but not least, my understanding of AI and business strategy has shifted tremendously. You don’t even need to love AI or technology. Maybe you hate it, as I do. But if you have a problem, you are able to develop a product.
I could verify the above on the Demo day.
It was a wonderful day to learn from each other. Other teams gave us feedback on how we could scale the product effectively and justify the costs. No one cared which websites I took the number from, but they were more curious about what we would do in scenarios A or B… So might as well just say the number is from my brain, and how about you get a cool suggestion from other people the next day. When listening to other groups, I noticed they also made a lot of assumptions rather than relying on a fixed piece of information.
Apart from my group, I love the learning platform for students and teachers to study Physics and Biology – I forgot the group number, but bravo, their prototype made with Loveable was so cool. The logistics platform to predict stock demands for fashion brands is fire, but I also forgot their group number…
