← Back to home

Case Study · Indeed · Big Easy

Evolving the design system for conversational AI products

Adapting a large-scale enterprise design system to support conversational workflows, generative interfaces, and AI-assisted product experiences across hiring and recruitment platforms.

AI product systems Conversational UI Generative interfaces Workflow architecture Design system evolution

Traditional product systems were not built for AI workflows

Indeed’s existing design system was originally designed around more deterministic product experiences:

As AI-assisted products expanded across:

the interaction model began changing significantly.

The new products introduced:

Traditional UI patterns were no longer sufficient on their own.

The challenge became designing a flexible system architecture capable of supporting products that continuously evolve as AI capabilities change.

Building systems that could adapt with the product

One of the core challenges in AI product design is that workflows are rarely stable early on.

Interaction patterns change rapidly as:

Rather than building rigid one-off interfaces, the system evolved toward flexible interaction primitives capable of supporting multiple workflow patterns.

The focus shifted from fixed screens to adaptable interaction systems.

From

  • Fixed screens

To

  • Adaptable interaction systems

This required designing:

The goal was creating infrastructure that could evolve alongside the product rather than needing to be rebuilt every time workflows changed.

Moving beyond static interface patterns

The conversational model introduced interaction behaviors that traditional enterprise systems were not designed to handle.

This included:

The system introduced reusable conversational primitives that allowed teams to build AI-assisted workflows without creating fragmented experiences across products.

These patterns supported:

The goal was consistency without constraining product experimentation.

Supporting dynamic interface generation

As products evolved, interfaces increasingly became partially generated rather than fully predetermined.

This required the system to support:

The architecture focused on composability and semantic structure rather than rigid page templates.

This allowed products to:

without fragmenting the broader product ecosystem.

Combining conversational and operational systems

One of the most important challenges was integrating conversational AI into real operational workflows.

Recruiters and employers still needed:

The system balanced:

Conversational flexibility with enterprise workflow reliability.

This included designing patterns for:

The result was not simply “AI chat,” but operational AI workflows embedded directly into employer product systems.

Expanding the role of the system itself

The work fundamentally changed the role of the design system.

From

  • Component distribution infrastructure

To

  • Interaction architecture for AI-assisted products

This required new thinking around:

The system increasingly became a platform for product evolution rather than a static set of reusable UI assets.

Enabling scalable AI product development

The conversational system architecture enabled multiple employer-facing AI products to evolve within a shared operational framework while preserving consistency, accessibility, and implementation scalability.

The system supported:

Most importantly, teams could iterate on rapidly evolving AI workflows without rebuilding foundational interaction infrastructure each time the product direction shifted.

Designing systems for adaptive products

The work established foundational patterns for how large-scale enterprise systems can support AI-native product evolution.

Rather than treating AI interfaces as isolated features, the system introduced a scalable interaction architecture capable of supporting:

The result was a more flexible product foundation capable of adapting as both AI capabilities and user expectations continued to evolve.

Read more case studies