Incident monitoring platform for autonomous markets — from a 100% manual process to a system serving 10 operators across more than 80 units.
01 — Context
Autonomous markets operate with open access — the customer walks in, picks products, and pays. But when something goes wrong — an incomplete payment, a theft, a repeat offense — someone needs to log it, investigate, and act.
All of that responsibility fell on a single person. They received the alerts, checked the spreadsheets, looked up the user's data, drafted the message, waited for a response, and updated the record. All manually, all in sequence, all dependent on their availability.
With more than 80 active stores, this model was at its limit. Any absence became a crisis. Any increase in volume ground the entire process to a halt.
02 — Process
Being a 0-to-1 project — no prior product, no design system, no internal reference — the process was built in layers: first understand the real workflow, then design, then build.
03 — Stack
Every technical choice was guided by one premise: delivery speed without sacrificing scalability. A lean, well-known stack with managed infrastructure.
04 — Product
Screenshot — Incidents screen
05 — Impact
Deliverables
06 — Learnings
The real problem is rarely the stated problem. The client thought they needed a better spreadsheet. What they actually needed was to eliminate the dependency on a single person — which is an architecture question, not a format question.
Designing under pressure is a design skill. Interfaces for operators in high-stress situations demand radical visual hierarchy — what matters right now needs to appear right now, without friction.
Design and development together accelerate everything. Handling both ends of the project eliminated handoff noise and allowed real-time adjustments — the interface evolved because I knew the true cost of every change.
A product in production is different from a product in Figma. The most valuable iterations came from real usage. What seemed obvious in the prototype was sometimes invisible in practice — and vice versa.