Work

Six case studies, start to finish.

AI platforms, investment banking and logistics. Each follows one structure: problem, constraints, decisions and trade-offs, solution, impact, and what I'd do differently.

01AI platform · private capital markets2026

Where AI lives in an institutional investment platform

AI left its separate chat page for reports, documents and meeting notes, in two modes: analysis to ask, edit to change, with review before saving.

AI-native Composer Design-to-code

Read the case study

02Loadsmart2021—23

Capacity matching for private truck fleets

A design sprint produced a platform. Six people from fleet teams tested it against a plain email and chose the email. We shipped it: response rate up 250%.

Research Product strategy

Read the case study

03AI platform · private capital markets2026

Specifying an AI composer, frame by frame

One bar held every AI feature, all shown at once. I emptied its first layer and respecified it: 26 frames, six state families, each with a line an engineer can pass or fail.

Interaction design State specification Prototyping

Read the case study

04AI platform · private capital markets2026—

A design system AI can build from

Most new UI came from AI tools. One canonical source, eight rule families so AI reuses what exists, and a register that sends anything it invents to review.

Design systems Rules layer Governance

Read the case study

05BTG Pactual2019—21

An advisors platform for four business verticals

A configurable dashboard, a modular side menu and a client workspace: the platform four business verticals finally agreed on. 12 interviews, 21 concept features, 90% acceptance.

Product strategy 0→1 Team leadership

Read the case study

06BTG Pactual2020

Modular card insurance in a banking app

Customisable coverage and assistance for a credit card. 80% of customers wanted to modularise their package. Usability testing moved the flow out of card signup to protect conversion.

Usability testing Complex flows

Read the case study