By Xante Philips – 648210
From Black and White to Blue and Yellow: What an IKEA Project Taught Me About AI
Where it started
When I moved to Amsterdam, I had to furnish a small room full of natural light. I could see its potential, but I didn’t know where to start. On the IKEA website, I discovered IKEA Kreativ, which let me visualize products I had already picked in a scan of my own room. It helped me see fit and style, but I still had to choose, combine and budget everything myself. So when our team set out to help IKEA customers with exactly that problem, I knew from experience why it mattered.
I didn’t expect much from AI. The few times I had brainstormed with ChatGPT, its answers felt boring and superficial. Not like me. I love creativity, music and color, and I’m outgoing and enthusiastic. ChatGPT never reflected any of that, and it kept misunderstanding what I meant. Could AI turn out to be more than the simple black-and-white brainstorm partner I remembered?
Role and contribution: a planner and a bit of a control freak
In the first week, I set up our first team meeting, created our shared document and drafted the planning. I’m a huge planner and a bit of a control freak: I always want to know what’s coming, so setting up an action plan is where I feel at home. Together, we set ourselves interim deadlines, so we never felt rushed in the run-up to our proposal, and submitted it on time.
I also brought my creativity into the project. I designed the test forms, the logo and the slides, using tips from my strategy consulting minor and IKEA’s exact color codes to make our presentation instantly recognizable. Beyond design, I created our ecosystem map and weighed the possible solutions against our success criteria, which helped us choose our preferred solution.
I enjoy group projects because other people’s insights inspire me, but trusting them is harder (the downside of being a control freak). When one of my teammates built our prototype, I had to let go of control and trust him to do it well. He worked hard, and it paid off. Relying on others turned out to be a relief, because doing everything myself puts a lot of pressure on me.
What surprised us: finding our gap, twice
Ironically, our first idea turned out to be too close to Kreativ, the very tool I had used myself. I was disappointed. We went back to the drawing board and kept narrowing our gap. In the end, the problem wasn’t that IKEA offers too much choice, but that customers have to turn a huge assortment into one complete room themselves.
Once we had that new direction, one of my teammates quickly built the first prototype. Then came the second blow: IKEA already offered something similar in the US. It just didn’t show current stock and wasn’t available in other countries, so we sharpened our gap once more.
This time, I took it much better. During a class break, I had asked our lecturer about our scope and first ideas. She told us it didn’t matter if something already existed, as long as we described exactly which choices we made and why. So instead of feeling defeated, we used it as motivation to research everything as precisely as possible.
We could have switched to another company or come up with a completely new idea. But choosing and combining furniture is a problem all of us had run into ourselves, and that personal connection kept us going.
Important decisions and iterations: putting a price on AI
With our gap finally sharp, the next decision was the price. Should AI room planning be free? We decided it shouldn’t. Every session costs money in AI usage, and IKEA already offers personal interior advice from €29.99 per hour. When I made our ecosystem map, I placed those human advisors as a complement to our tool, not a competitor. A free tool could push them aside, and their interests count too.
So we built a pricing ladder, which in class is called versioning: free self-service tools, our AI planner at €4.99 per room, and human advice from €29.99 per hour. At €4.99, personal advice becomes accessible to people who would never pay €29.99 an hour. One of my teammates suggested crediting the €4.99 back when customers buy the furniture. So if they buy the room, the advice is effectively free.
Feedback and evidence: learning from each other
For me, learning together is the most effective way to learn, and our final weeks were full of it. We compared our prototype with IKEA’s current website: each of us tested four participants, sixteen in total. Saving time was one of our goals, and our prototype cut the median planning time by more than half. But only half of the participants would pay €4.99, even if it were credited back on purchase, so our pricing is still up for discussion.
At Demo Day, I presented our project with my three teammates, and we assessed the other groups’ work. People were curious how we had built our prototype and impressed by our sixteen participants. They suggested a 360-degree view and a floor plan from above, which matched our test: four participants had missed or doubted the room view. And because we presented several times, each round taught us to explain it better. This blog, Demo Day and the Course Learning Reflection made my learning fun and more conscious.
Collaboration: working together, apart
All those changes in direction tested us as a team. What worked best was our fast communication. We kept in touch through WhatsApp and worked at the same time, so we could change course quickly and help each other when someone got stuck.
What didn’t work was the waiting. Sometimes only one or two of us could move forward while the rest waited, so we still ended up working on weekends. We live in different cities, so much of our work was online. Calls were short and practical, and afterwards everyone worked alone. I missed sparring over coffee, where the creative ideas come from.
Next time, I want us to plan further ahead as a team, not just two steps at a time. That gives ideas time to rest, so new insights can surface.
Changed understanding of AI and business strategy: not new, but ours
Before this project, I thought a valuable AI idea had to be new. Now I see it differently. Our idea already existed. What made it valuable was how sharply we defined it, and for whom. One AI tool affects customers, IKEA and its employees at once, so it should be more than an efficiency tool for a company. My advice to any team that finds their idea already exists: don’t start over, but describe exactly which choices you make and why.
I also learned that AI rarely gives the best answer straight away. Our prototype took round after round: write the prompts, test the result, then use the feedback to adjust them. So was AI really that boring? Maybe it just needed someone to bring the color.
