SeamlessAI

Context

In early 2025, SeamlessHR moved from traditional SaaS toward an AI-enabled platform — embedding intelligence into existing enterprise workflows across recruitment, HR, and operations rather than shipping AI as a standalone product. With multiple product teams building AI features independently, the company needed a shared design foundation for how AI would surface decisions, handle errors, and earn trust in high-stakes enterprise contexts.

Problem

AI capability without a design language to support it. Teams were building AI features independently — different patterns for surfacing recommendations, different error states, different levels of explainability. Users encountered AI behaviour that varied across the suite, which created unpredictability in high-trust workflows like recruitment and payroll. The inconsistency didn't just look wrong; it eroded user confidence in the features themselves.

Approach

Partnered with AI/ML, product, and engineering to define an AI-native UX layer across the SeamlessHR suite. Started with recruitment — where AI impact was most measurable — and worked backwards to a reusable pattern library. Defined prompt frameworks, agent behaviours, and interaction models that all product teams could implement consistently. Scoped analytics and reporting with the ML team to ensure the interface reflected what the model could actually deliver, not aspirational states the pipeline couldn't support. Designed Smart CV Parsing, Smart Ranking, and Smart Assessment flows as the first system-native AI features.

Outcomes

AI features across SeamlessHR moved from individually designed one-offs to a governed system of reusable interaction patterns. Smart CV Parsing, Smart Ranking, and Smart Assessment — each built on the shared layer — reduced screening time while making the system's reasoning visible to recruiters. The pattern library became the foundation other product teams adopted when building their own AI features, compressing design time for each subsequent release. SeamlessHR now has an AI design system, not just AI features.