03 · Logistics
An eight-step form, replaced by a conversation
I redesigned scenario creation around intent, while keeping pricing and operational constraints inside a deterministic engine.
Working prototype 03 Logistics
01 · Context and ambiguity
The useful problem was underneath the brief.
Users had to translate a business scenario into an eight-step configuration flow before the product could help them evaluate it.
02 · Decisions
I made the system explicit before polishing the surface.
- I let conversation capture intent without asking users to learn the data model.
- I kept calculations and constraint resolution outside the language model.
- I made assumptions editable before a scenario could be committed.
03 · Human and AI boundary
Automation got a lane. Judgment kept ownership.
The LLM interpreted intent and narrated results. A deterministic rules engine owned pricing, validation, and constraint resolution.
04 · Working prototype
The artifact became the research surface.
The prototype connected conversational input to structured scenario data, visible assumptions, and comparable outcomes.
05 · Outcome
What changed
The team could evaluate conversational UX without compromising the reliability of the underlying operational model.
The system boundary was tested in a functional prototype rather than a static chat mockup.