Industry Occupation
Insights.

I led UX on this project end to end over 12 months, from the initial proofs of concept through problem reframing, the modular system design, and validation with public users and coaches. The work spanned design research (interviews, surveys, usability validation), interaction and information design for the signal framework and dashboard, and close collaboration with WSG stakeholders through to an accelerated rollout decision.

Industry Occupation Insights (I/O Insights) closed that gap. It is a modular, signal-based dashboard that gives coaches real-time industry context, including demand, wage movement, growth outlook, and career pathways, assembled per session and shareable live with clients. Coaches reported a 62% reduction in research time per client, from 27 minutes down to 10, and WSG's ACE division requested an accelerated rollout, a rare outcome for a government technology project.

The project sits within WSG's broader career ecosystem, anchored by four core pillars. These are not abstractions. They set the boundaries for every design decision: understand behaviour before prescribing action, make opportunity visible and measurable, build tools that adapt to shifting futures, and design for an ecosystem rather than a single platform.

Each pillar translates into a concrete strategy. Behavioural Intelligence shaped how the dashboard models coach decision-making. The CORE engine defined the metrics layer. The Career Guidance Engine became the I/O Insights product itself, mapping skills, jobs, and pathways into actionable views. The Ecosystem Integration Hub ensured the system could connect to partners and scale beyond a single use case.

Research time has dropped substantially with the use of I/O, yet most users perceive no change, suggesting the time savings are not yet consistently felt.

Most coaches find the tool moderately useful, consistent with its genuine but situational value. It is used selectively for specific client profiles and coaching contexts rather than as a routine resource.

The project established the foundation for the broader Career Portal initiative, extending the approach toward policymakers, employers, and educational institutions.

The thing I carry forward from this project is not the metric, it's the mechanism behind it.

I/O Insights minimised effort and maximised the coach's sense of control at the same time, and it did so specifically because construction stayed in human hands. The survey data tells a subtler story: research time dropped 62%, but most coaches didn't perceive the change, and usefulness rated 2.75 out of 5. These are not contradictions. They are evidence that the tool's value is situational, not universal, and that the perception of usefulness lags behind the reality of time savings. Trust didn't come from the intelligence of the system. It came from the legibility of it: coaches could see where every signal originated and decide what to do with it.

That observation is what pulled me toward research on human agency in AI-mediated tools: the question of how much a system can do for someone before it starts doing it to them.

The shipped product sits at the manual, human-controlled end of a much larger design space. The interesting research question is whether newer LLM and agentic capabilities can take on more of the construction work without spending the trust the manual version earned.