The Retail Problem That Sparked Zenline
You’d think global retailers know what’s on their shelves. After all, these are billion-dollar organizations with access to data, tools, and talent. But during our first years working at retailers and consulting firms, we kept seeing the same thing: Even the most sophisticated retail organizations struggled to answer basic questions like: Which products are driving margins? Which categories are bloated? How do we compare to our competitors?
This disconnect was eye-opening. Strategic assortment decisions were often based on gut feeling or outdated dashboards rather than real performance data. Millions worth of inventory sat idle while pricing strategies lagged behind the market. This wasn’t just inefficient, it was a huge missed opportunity. All while global platforms like Amazon and Temu were setting new standards for speed, responsiveness, and data-driven precision.
The world is changing, so should retail
AI is no longer hype. Every industry is being reshaped by intelligent software. Our vision: A future where every retailer has AI agents that handle analysis, flag cannibalization effects, identify assortment gaps, and recommend actions – while humans focus on execution, negotiation, and decision-making. Category managers will evolve with and around AI, instead of getting replaced.
Our journey began with fast iteration
For AI in retail, it felt like we were the perfect match: Arber with a background in strategy and consulting at BCG, Gerrit a seasoned engineer with experience building AI systems with experience from Amazon. Years of friendship and joint projects at ETH Zurich created a foundation of trust, shared ambition, and speed. In a landscape where speed is everything.
When we started building Zenline, we explored both sides of the market: manufacturers and retailers. Initially, we weren’t sure where the bigger pain point was. But after talking to dozens of category managers and validating assumptions through prototypes, one thing became clear: the real pull came from retailers. It’s where the problem was most urgent and where decision-makers were actively looking for tools that actually worked.
The first pilots came through warm intros, the next ones (with more confidence) from outbound. We listened, kept iterating, and quickly saw that the value was in giving category teams better assortment decisions.
Thanks to many, many, many 5am nights, and a lot of AI usage, we were able to build our first version within four weeks and went live.
The most telling moment came right at the start
Amidst all the late-night prototyping and fast feedback loops, there was one moment that stuck with us: A surprise launch party, thrown by a group of close friends – many of them engineers, investors, smart people – with one message: “We’re behind you. Even if you never answer our WhatsApp messages.” We had people around us who believed in the vision and pushed us to go all in. That’s when we knew: this is real. Now it’s full speed ahead.
