AI decision tools for a ports operator
Research with ship captains that uncovered why vessels waste fuel waiting — and a set of AI use cases designed around the answer.
Client work, shown abstractly. Commercial port operations are sensitive, so there are no real screens, routes, or pricing here — only the research, the insight, and what I designed around it.
A ports and logistics group wanted AI to make its maritime operations smarter: help captains sail at the right speed, forecast where empty containers need to be, coordinate berthing. The question wasn’t “can we build AI” — it was “AI for which decision, made by whom, that’s actually broken today.”
Going to the source
To answer that, I did UX research with people most software teams never talk to: working container-ship captains. When direct access was hard, I recruited through maritime forums. What came back reframed the whole program.
Captains rarely get told a speed — they’re told “maintain ETA” and left to guess. And the reason ports don’t share berth availability isn’t an oversight: keeping that information opaque helps keep terminals full. The fuel waste and idle anchorage time everyone wanted to fix were sitting on top of an incentive problem, not just a data problem. Naming that out loud changed which use cases were worth building.
What I did
- Ran the captain research and synthesized the commercial-team workshops into a current-state map of how voyages and berthing actually work.
- Designed the priority AI use cases — voyage speed intelligence, container repositioning — as exec-ready one-pagers: the problem, the users, what the tool does, the decision it improves, the business value.
- Designed the interface for the container-repositioning tool, including the decision drawer where a planner sees the recommendation and the reasoning behind it.
- Drove the Phase-2 use-case catalog — scoring ideas on desirability, viability, and feasibility, and insisting every one traced back to a real user and a real friction.
Outcome
The captain insight reset the program’s priorities. Use cases that looked good on paper but sat on top of an incentive problem were cut before they consumed budget, and the voyage-speed and container-repositioning tools moved forward as exec-ready one-pagers — each with a clear decision, user, and owner behind it. Phase 2 ran on a use-case catalog scored the same way, so the group had a defensible roadmap instead of a wish list.
Why it matters to me
The strongest thing I brought wasn’t a screen — it was an insight from talking to the right people, and the discipline to let it kill the use cases that wouldn’t have worked. Executives don’t buy dashboards; they buy better decisions. This project is how I learned to design for that.