Hi there, welcome to my project learning blog for the course BM01BIM Information Strategy! I would like to walk you through my five biggest learning milestones of this block, so you can get a better understanding of my experience during the AI Strategy Lab project.
Lesson learnt 1
This was my first group project in my MSc programme, so it was really my first exposure to and impression of the cohort. The group dynamic turned out to be one of the more interesting parts of group projects, as it was notably different from my BSc experience.
Our team of four came from different bachelor’s degrees, different cultures, different amounts of work experience, and different motives for choosing Business Information Management. Coming from IBA where group work meant peers at the same stage of the same degree, it was immediately obvious how different this was: people already carried real professional habits and instincts into how they approached the problem, disagreed, and divided work. This difference shaped the whole project more than I expected going in, and it’s probably my first lesson: don’t assume the group will work like your bachelor’s groups did, because the people in the room have already learned to work in very different ways.
Lesson learnt 2
Within this dynamic, I ended up taking on a de facto co-lead role with one of my teammates. He was the owner of the idea we ended up developing and had the most background experience and knowledge on the topic, so he was responsible for the theoretical checks. Myself, I took care of the administrative leadership tasks: I initiated and led most of our meetings, for which I always prepared in advance to not let it turn into an open-ended discussion going on for hours and hours, as we all had busy schedules working part-time jobs or internships next to BIM. I gave constructive feedback on teammates’ sections as we went, and in turn learned a lot from watching how the others approached their parts, particularly where their professional backgrounds showed through in how they structured arguments or pushed back on vague claims. After being inspired by one of my teammates who openly approached the professor for redirecting feedback on our idea, I also took on the courage and approached the TA to clarify some assignment requirements and make sure we would deliver the project according to the deadlines. I also kept meticulous track of everything said and discussed in a shared doc, which did not just come useful to serve as a central database, but is also now helping all teammates write up their individual assignments, with the logged timestamps and decisions. This was another important lesson: co-leading a group.
Lesson learnt 3
My main content contribution was developing the Monetize workstream, including the pricing model, which is also one of my key takeaways from this course, as I had to become very familiar with this content and even conduct further research to ensure I had a perfect understanding of the topic. This is the part of the project where I iterated the most, by a clear margin. I came up with multiple pricing models before finally feeling satisfied with the final product. First, I followed my previous encounters with pricing strategies and developed two models that are common in the real world but were not perfect for our agentic layer. In the meantime, we also had a minor pivot with our original idea, so just when I thought I was finished, I had to start again from scratch. I rebuilt the entire model from scratch on top of actual published benchmarks instead (AI agent build costs, enterprise SaaS pricing ranges, comparable inventory-forecasting tools) and went through several more versions after that before I was satisfied: first getting the benchmarks right, then converting to EUR, then building a complexity-weighted formula instead of a flat per-station number, then stress-testing it against two deliberately weaker alternatives (a flat fee, a pure usage-based fee) to make sure I could defend why ours was better and not just different. The final version is still an estimate, but it’s one I can trace back to a source for every number in it, which the first draft never could have survived being asked about.
Lesson learnt 4
To expand on the pivot briefly mentioned above, indeed, at one point we learnt that our idea was not novel enough and we had to re-scope mid-project. It meant narrowing from the entire forecasting-and-ordering pipeline down to one specific decision point: the gate that decides whether an order auto-sends or escalates to a human. It was frustrating, but it taught me another important lesson: a narrow, well-evidenced claim beats a broad, impressive-sounding one almost every time, and that’s true for pricing models as much as it’s true for the overall pitch.
Lesson learnt 5
Throughout this project, my understanding of AI strategy also changed. When starting the course Information Strategy, I had a fairly blurry sense that AI strategy meant describing how a company could use AI and listing the benefits. I’m leaving with a much sharper sense that the actual work is almost entirely about defensibility: where the real boundary of innovation sits, what can be sourced versus what’s just asserted, and what a reasonable executive would actually ask you to justify before approving a budget line. Another takeaway of applied AI strategies was about scope and claims. The pivot after Ting’s feedback taught me that AI strategy claims fail or succeed on precision. “We automated replenishment” sounds impressive but can’t survive “a vendor already does this.” “We made one specific decision smarter, with a clear baseline” is less exciting, but it’s a claim you can actually test and defend. I’ll carry that instinct forward: ask what the smallest true claim is before reaching for the biggest-sounding one.
Overall, it was a very eventful project and course with many learnings, both academically and personally.
