QChat

The challenge

Queenslanders were already using consumer AI at work. The problem was they were doing it unsafely, across tools that weren't built for government data.

QChat existed to change that. My job was to make it the tool people actually chose.

‍

‍

Designing for everyone

A finding that only held for half the spectrum wasn't a real finding.

Seven personas. Daily AI power users on one end. First-time users on the other. No designing for the middle.
Every decision had to work at both ends simultaneously.

‍

‍

Discovery over onboarding

Users already knew how to query. They'd been doing it for years.

The interface didn't need to teach behaviour. It needed to show what was possible. Prompt suggestions surfaced as invitations rather than tutorials. Progressive actions appeared as conversations deepened, drip-feeding capability without overwhelming new users or boring experienced ones.
‍

‍

‍

Data security

A trust problem disguised as a labelling problem

Data security classifications in QChat mirrored the QGEA framework directly: Official, Sensitive, Protected.
That turned out to be the problem.

Protected existed in the framework. It didn't exist in QChat. Users could select it, but nothing changed. Testing with 8 participants confirmed it: the interaction was described as illogical and paradoxical.

Offering a security setting that implied a capability the product couldn't deliver wasn't a UX problem. It was a trust problem.

The redesign started with honesty. Two modes, not three. Official as the frictionless default. A Sensitive toggle for data that shouldn't be retained, with web search auto-disabled and a persistent mode indicator throughout the conversation. Protected was removed entirely.

An option that creates false confidence in a security context is worse than no option at all.

‍
‍

‍

Projects

Projects was the most complex problem on the product. A conversation organiser that was also a knowledge base, where uploaded documents shaped everything within it.

It split into two types: user-created and admin-curated. This meant designing for authorship, trust, and visual hierarchy simultaneously.

Then the terminology problem. "Project" carries real operational weight in Queensland Government. PRINCE2 and Agile frameworks use it precisely.

Getting it wrong wasn't just confusing. It was a data security risk.

Pre-made became "Pre-made Projects." Custom became "My Projects." Shape and colour carried the state change between starting point and owned asset. Author attribution established trust and accountability.

‍

‍

Outcome

AI adoption at this scale doesn't require technical literacy. It requires design that earns trust incrementally.

QChat is now proof that government technology can be both secure and genuinely usable.

The response components and conversational UI patterns built for QChat are being carried into other Queensland Government AI products.

‍