0→1 product design
The Hen House: Data visualisation
Redesigning a broken, unsaveable submission form into a structured wizard, helping Berlin's digitalisation network, funded by BMWK, finally capture and share the project knowledge it had been losing to drop-offs.
Role
Lead Designer
UX Architect
Platform
Desktop (manager) +
Mobile (driver)
Deliverables
MVP +
documented roadmap
Industry
Operations / AgriTech logistics
Duration
8 months,
Dec 2023 - July 2024
Team
PRODUCT And THE PROBLEM
Disconnected systems,
invisible information,
zero structure
Acme Today is an animal rendering facility with a workforce of managers, drivers, vehicles, and daily tours to coordinate. The reality: multiple spreadsheets nobody trusted, couple of WhatsApp groups for last-minute changes, and phone calls to figure out who was available. No single system held the truth.
The need was a platform unifying two user groups: managers on desktop planning and optimising week-ahead tours, and drivers on mobile receiving assignments, requesting leave, and staying informed in real time.
AI was introduced at the points where it genuinely reduces load, specifically the cascading replacement logic where one absent driver triggered a manual chain of calls, checks, and re-assignments under time pressure.
business need
Hours, availability, and experience are tracked manually, often incorrectly. Compliance risk is baked in by default.
business need
No operational visibility
No centralised data meant no way to track efficiency, spot bottlenecks, or plan ahead.
user need
Fragile replacements
One absent driver triggered a manual chain of calls and re-assignments , under time pressure, with no system to catch it.
user need
Drivers kept in the dark
Tour assignments, approvals, and petrol pins all via phone call - no confirmation, no record.
ROLE AND CONTRIBUTIONS
Sole designer across both platforms, from discovery through to developer handoff. Worked directly alongside the developer to co-design the AI integration points.
Mapped every operational constraint, working hours, rest periods, route experience, and shift-lead rules before drawing a single frame. Mapped the full replacement flow across states, including cascading edge cases nobody had formally documented.
Designed the dual-sided Information Architecture.
Manager desktop covering tour planning, CRUD (create, read, update, delete) management, leave approvals, and the AI suggestion panel.
Driver mobile covering tour notifications, leave requests, and operational messages, including petrol pin.
Took both from paper prototype through mid-fidelity wireframes to high-fidelity, with annotated handoff documentation covering MVP scope.
KEY DESIGN DECISIONS
decision 01
Map the replacement flow before designing a single screen
The replacement scenario was the hardest problem and the most important to solve first. Mapped the full cascading flow, including edge cases nobody had documented. That map became the skeleton for both the IA and AI integration.
decision 02
Designed for low-tech confidence, not just functionality
The primary users, managers averaging 50+, many with low-to-moderate tech comfort, needed a system that felt familiar, not clinical. Every interaction was stripped to its simplest form, plain-language labels, no hidden logic. The goal was confident onboarding, not a steep learning curve.
decision 03
AI as an assistant, constraint-aware, not intrusive
When a driver is marked absent, the AI checks every constraint before surfacing a suggestion: daily hour limit (10h), weekly overtime cap (5h), route experience, driver role (Expert / Trainee / Factory-fill), and holiday rules. Only valid candidates appear, ranked, not imposed. One clear suggestion, just enough context to decide. No noise. One tap to apply, full freedom to override.
Logic 01
The manager's weekly calendar view surfaces AI suggestions directly on the schedule, before a tour is manually assigned. The AI recommends the most suitable driver for each tour based on:
Availability and remaining working hours
Role match and route experience
Tour frequency, avoiding overloading one driver
Continuity across weekly tour cycles
The AI works in the background, the manager confirms, adjusts, or overrides. Planning shifts from reactive to informed.
Logic 02
When a driver is marked absent, a replacement pop-up triggers immediately. The AI resolves the entire chain reaction. The AI checks every candidate at every link in the chain against:
Hours worked + daily (10h) and weekly (5h) overtime limits
Driver role - Expert / Trainee / Factory-fillmatch per route
Route experience and tour history
Holiday rules
Only constraint-cleared candidates are suggested at each step. Apply or override, a human is always in control.
OUTCOME
The platform shipped as a working MVP, the company's first centralised workforce management and planning system. A single stop for tour planning, driver availability, scheduling conflicts, and workforce communication that had previously lived across disconnected tools.
For a branch managing 50-70 drivers, 10-20 trainee drivers, and 10- 15 in-plant alternates, the AI replacement flow replaced what had been an all-afternoon manual process, cross-checking multiple spreadsheets and recalculating clocked hours, with a single validated suggestion, every constraint checked before a candidate was shown.
Drivers received their first structured interface, tours in advance, leave requests in-app, and petrol pins delivered directly. Both groups described the change as removing a daily source of friction.
The platform was showcased at 11 tech open house events and exhibits, drawing interest from logistics firms, fleet managers, and scheduling coordinators, validating its adaptability beyond the original animal rendering context.
9
9
9
9
5
5
5
%
%
%
Reduction in replacement planning time
1
1
1
1
0
0
0
X
X
X
Faster Tour Planning Process
1
1
1
1
0
0
0
0
0
0
0
0
%
%
%
Centralised Workforce Planning and Management
