Innovation at Work: Prototyping for Impact
“Prototyping often sounds like something reserved for new products or startups. But in reality, it can be just as powerful inside an established manufacturing environment, especially when the goal is to make complex processes simpler and more efficient.”
In my first year of the enFocus fellowship, I had the opportunity to work with one of our sponsor organizations, an aluminium extrusion company. Their facility operated in a high-mix, low-volume environment, producing over 100,000 unique parts. That level of variety is impressive, but it also came with frequent changeovers and long lead times. The company was eager to move toward cell-based manufacturing, but to get there, they first needed to standardize production routes and better understand how different parts were related. That’s where my project came in: creating a lightweight, data-driven tool that could help engineers group similar parts based on their key attributes.
I started where most of us do, in Excel. It was fine for exploration, but it wasn’t built for automation or scale. So I moved to Python and began experimenting with clustering algorithms. They produced neat visualizations, but they also oversimplified the data and ignored how manufacturing decisions are actually made.
“Then, one day while standing at a grocery store checkout, I noticed the barcodes on the products and thought, “What if parts could have barcodes too?””
That idea became the heart of the tool. Each part had four main attributes: route, thickness, material, and length, but each of those had multiple categories. By converting these categories into binary (yes or no) indicators, the data expanded into fifteen variables. In simple terms, I turned descriptive attributes into a digital “fingerprint.” By combining these into a single ID, the tool could identify parts that were either identical or closely related.
The logic for this framework was refined through collaboration with the sponsor’s engineering team and project manager. Together, we agreed on how to rank the features by manufacturing importance, ensuring that the grouping made practical sense for real production decisions. This back-and-forth was invaluable, and it transformed the tool from a technical exercise into something that reflected the team’s actual needs.
We tested the prototype on a pilot set of 400 high-frequency and high-volume parts. When I first demoed it, some team members were hesitant about using Python scripts, but after seeing how quickly the tool delivered insights, their interest grew. With a clear standard operating procedure in place, the team began to see it as something they could actually use.
Time studies are still underway, but the early feedback has been encouraging. More than the technical side, the experience reinforced a few lessons for me. Prototyping helps clarify what truly adds value. Collaboration is key to making a tool usable. And sometimes, the best ideas appear in the most ordinary moments, like a barcode at a grocery store.
Looking ahead, there’s potential to take this further, perhaps by turning the Python script into a standalone application, integrating it with the company’s existing data systems, or using it as a foundation for more advanced analytics on production planning. Whatever direction it takes, the project opened up meaningful conversations about how lightweight digital tools can create real value on the shop floor.
In the end, prototyping for impact isn’t about building perfect solutions. It’s about making ideas tangible enough for others to see, test, and build upon, and that’s where real change begins.