A Year at Tradify
How an AI-forward product team let one designer own scope, code, and outcomes — and still ship measurable activation gains.
Industry
B2B SaaS
Duration
12 months
Role
UI UX Designer
Team
Cross-functional
workflow used to speed up research synthesis and engineering fixes
most-used actions surfaced directly on the job card
design system rebuild - tokens, variables, nested components
Overview
Tradify is a field service management platform used by trade businesses across New Zealand, Australia and the UK. For over a year I am working across product feature design, a growing design system, and an AI-assisted workflow to accelerate research synthesis and delivery.
It starts with AI
Tradify set out to become one of New Zealand's most AI-forward product companies. That meant giving designers and engineers access to real AI tools - and the trust to use them in day-to-day work. Here's how that changed the way I design.
Claude handles the work that doesn't require judgment - generating and testing label variants, auditing designs against system rules, and catching inconsistencies before they reach review. That reclaimed hours I could spend on product thinking, UX decisions, and collaboration.
Figma Make accelerated exploration. Instead of testing only a handful of concepts, I could quickly generate and iterate across dozens of viable directions before refining the strongest ideas. The result was stronger concepts, explored faster and with greater confidence.
BigQuery in Claude removed the guesswork. By querying live product data directly, I could base design decisions on actual user behaviour instead of assumptions or waiting for reports.
Then there's Devin. Because it understands our codebase, I can confidently work across the boundary between design and engineering. With Devin and GitHub, I often write and implement my own technical solutions, with engineers reviewing and approving the final changes. Ideas move from Figma into production without losing context along the way.
None of this replaces design judgment. It removes the mechanical work so I can spend more time understanding problems, making better decisions, and improving the product.
AI Assisted Design stack
Working in ShapeUp cycle example
Reducing churn had become Tradify’s highest priority customer loss was approaching the rate of new acquisitions. The analysis was pointed: 78% of churned customers never activated, despite 72% having already used both web and mobile. The problem wasn’t awareness, users were evaluating the product and leaving before ever feeling its value.
Research
The job list was the most-used screen, yet users had to open a job and move through multiple tabs to complete common tasks. A survey of 126 customers ranked the actions used most: navigate to site (53%), add notes (39%), upload photos (33%), and track time (33%).
Design approach
Instead of redesigning the job details screen, I brought the four most common actions directly onto each job card, turning a passive overview into an action-focused workspace. Layouts were shaped by production usage data, not assumptions, and functionality adapted by list context (Today, New, Active, All) and subscription tier.
Collaboration
One constraint surfaced: the timer could only be controlled from the screen where it started. Rather than design around it, I worked with engineers to understand the architecture and, with help from AI tooling, identified a simpler refactor. The team agreed, and we shipped a seamless cross-app timer.
Outcome
Twelve improvements shipped under a single A/B flag to measure activation impact, with activated customers (10+ jobs in 30 days) as the primary metric.
Early results showed a positive lift in activation versus control group, plus improved progression to paid in exposed cohorts.
Examples of deliverables in ShapeUp cycle
Mobile application before and after of Job List screen
The foundation underneath
Moving this fast only works with something solid underneath. Alongside the cycle work I led a full rebuild of Tradify’s design system on modern Figma tooling, variables, tokens, and nested components, so consistency is enforced by the system rather than remembered. It became the shared source of truth as Tradify moved to React, cutting design debt and easing onboarding for new designers and engineers.
The rebuilt Figma design system