Enterprise UI and AI ideation at a Big 4 firm.

Design system. AI features. Ten products in production.

Role Product Designer

Timeline 2024 - present

Team Multiple product teams

Status NDA · anonymized

Existing tools that need to grow, not just look better.

Internal enterprise software at a large professional services firm doesn't get replaced. It evolves.

The challenge isn't building from scratch. It's understanding what's already in use, where the UI creates friction, and which AI capabilities could fit inside existing workflows without breaking what already works.

Some projects needed clean UI updates. Others needed someone to think through what these products might look like in two years.

Stakeholder-led. PM-first.

Direct user access on this engagement is limited. Requirements come through product managers who work closely with the business.

My process is about translating those requirements accurately. I focus on understanding what the PM needs, asking the right questions, and pushing back when something won't work as UI.

For AI features, I designed concepts like an in-product assistant that answers plain-language questions about live data and returns structured tables with a recommendation. I designed it in light and dark from the same components, so it drops into existing products without a reskin.

An enterprise AI assistant concept shown side by side in light and dark mode: a chat thread summarizing Q3 regional performance, a data table of EMEA, Americas, and APAC revenue with QoQ change, a flagged recommendation, and quick-action buttons. — AI assistant concept. Plain-language query, a structured result table, and a flagged recommendation, designed in light and dark from one component set. Anonymized.

Design system, dark mode, AI features.

I was identified early as the team's data visualization specialist. A significant part of my work has been contributing to the firm's shared design system, specifically converting light-mode components to dark mode.

That means maintaining consistency while making decisions that hold up across different screen contexts.

Alongside that, I've designed AI features. I think through how to embed intelligent capabilities into tools that already have users and established habits. This isn't concept work for a pitch. It has to fit into products that are running.

I also worked on UI improvements across multiple development teams.

Shared component library shown in dark and light mode side by side: primary and ghost buttons in rounded and square variants across large/medium/small sizes, closed and open dropdowns, toggle switches, and AI feature chips in active and inactive states. — The shared component set in dark and light. Buttons, dropdowns, toggles, and AI feature chips, each spec'd at parity across both modes from one source of truth. Anonymized.

Dark mode components shipped. AI features in development.

Components I designed are part of the live design system. AI feature designs have moved into active development discussions. The engagement is ongoing.

What I'd do differently.