Designing the smart pantry app
Busy households run out of essentials because keeping track of them is invisible mental work that competes with everyday life.
Blum — a world leader in kitchen hardware — partnered with Bridgemaker to build a product that takes this burden away: Kiip, a smart replenishment app that learns what your household needs and reorders it before you run out.
I led the end-to-end design of Kiip from 0 to 1 — research, prototyping, UI and brand — as the sole designer in a cross-functional venture team. Kiip launched in late 2023 in partnership with REWE, reaching 2,200+ registrations in its first three months.
Role
Lead Product Designer
Platforms
Web App
YEAR
2024
Team
1 Product Manager
6 Backend Engineers
2 Frontend Engineers
1 Data Scientist
1 Marketing Manager
Hello David! I'm Kiipy, your intelligent inventory assistant 👋
I monitor your supplies and make sure you always get restocked in time.
The more I learn about you, the better suggestions I can make for you.
The Challenge
An algorithm can only predict consumption if users feed it data — but users only engage if the app feels effortless from day one. We needed to:
Make onboarding feel like meeting an assistant, not configuring a tool — collecting only the minimum needed to start
Balance user control with automation, so predictions stay accurate without asking anyone to micromanage a pantry
Build trust between the first and second order, when the algorithm is still learning and its value isn't visible yet
The Approach
1. Automate the restock decision
Usability testing showed users valued simplicity over granular control — so we designed the system to decide by default. Kiip's algorithm calculates the usage of each product, predicts when it will run out, and adds it to the next delivery automatically. Every delivery stays fully editable: the user confirms, adjusts or removes with one tap.
2. Let Kiipy do the heavy lifting
Setup collects only two things: household size and five starter products. Kiipy, the app's conversational assistant, walks users through it in under a minute — making the first session feel like meeting someone helpful, not filling in a form. Everything else is learned over time.
3. Build trust with contextual guidance
Handing your household to an algorithm is a big ask. Instead of front-loading explanations, we placed guidance cards at the exact moments users meet each feature for the first time — predictions, quantities, deliveries. And one line of copy did more work than any feature: “kiip ist kein Abo.” Users keep full control over every delivery.
Impact
Market validation
3,000 households signed up in the first month on a lean, Instagram-only budget — enough signal for Blum to keep investing in the venture.
Onboarding completion rate
45% of registered users completed their first manual order and entered the consumption-tracking cycle.
Usability testing task success rate
81% of interview participants successfully completed core tasks (setting up an order, understanding how automated restocking works) without assistance.
Looking back
Kiip asked users to hand over something personal — their household — to an algorithm. Working on it taught me that trust isn't a feature you add at the end: it's built in every line of copy, every guide, every confirmation state. The best product is the one users never have to think about — and designing that invisibility is the most rewarding work I've done.