Field Notes · Elle Anderson · andersonco.ukField note · Case study ยท AI for good
Find My Brick: Giving Creativity Back, One Missing Piece at a Time
Between 2017 and 2020 I was Product Director for data science at The LEGO Group, and I had the privilege of leading Find My Brick from its first conversations to its growth. It began as an experiment with artificial intelligence, but the problem we carried into it was a human one, and it was close to our hearts.
What Was on Our Hearts
LEGO exists to inspire and develop the builders of tomorrow. Yet some people were saying, not unreasonably, that the modern kit, with its instructions and its single intended model, had taken something away from children: the free, open-ended building that had shaped generations. We felt a real commitment to flexing those creative muscles again, and to showing that technology could serve that purpose rather than compete with it.
At the same time there was a small, everyday sorrow we all recognised. A child reaches the last page of the instructions and one piece is missing. The build stops. The parent faces a parts list, a tiny component they cannot name and a wait. We wanted to turn that moment of disappointment into a moment of delight, and we wanted to use it to open a door to building anything at all.
What We Built
Find My Brick let anyone photograph a brick and have it recognised, matched to the catalogue and linked to a replacement and to LEGO.com. The vision model was built by a data science team doing wonderful work. My responsibility was the product around it: the experience of the child and the parent, what happened when the model was unsure, how we earned trust one identification at a time, and how we reached families around the world. Working closely with marketing, we grew it to two million users, and the experience sustained a Net Promoter Score of 90 to 95.
What It Taught Us
Begin with a human purpose. We did not set out to prove that we could use artificial intelligence. We set out to give something back to children, and the technology earned its place by serving that. Every decision became easier once the purpose was clear, because the question was never "what can the model do" but "what does this child need next".
The experience around the model is where the care goes. The model was a gift from brilliant colleagues. The product was everything around it: what a child saw when the match was uncertain, how gently the experience recovered from a mistake, and how quickly a family got back to building. That is where trust was made.
Reach is an act of generosity. Growing to two million users was patient work with marketing, store teams and the community. It mattered because every family we reached was a family whose build did not stop, and a child who might carry on building something of their own.
You can design for trust. We measured accuracy as a family experienced it, not as a laboratory reported it, and we treated each wrong answer as a promise broken. That discipline is what let the product earn its place in people's homes.
What Stays with Me
I still think of Find My Brick as the clearest example I have of technology in the service of something better: a child, a missing piece and a door opening on to everything they might build next. It is the standard I bring to every AI product I have worked on since.