IBM Watson Conversation
I was designing conversational AI tooling in 2016 — when most people still called it "chatbots" and meant it dismissively. That's not a resume detail. It's a decade of compounding pattern recognition.
- Role
- Visual Design Lead / UX Designer
- Company
- IBM
- Period
- 2016–2018
IBM was building the tooling layer for enterprise conversational AI at a time when the technology was nascent and the market barely existed. Watson Conversation (later Watson Assistant) needed to empower non-technical users to build chatbots and virtual agents driven by natural language understanding. The challenge was making genuinely complex AI configuration — intents, entities, dialog trees, NLU training — accessible to business users without dumbing it down for technical ones.
Conversational AI had no established design patterns. We were inventing the interaction model as the technology matured beneath us. The platform had to serve both developers (who thought in APIs and JSON) and business users (who thought in customer journeys). IBM's enterprise customer base demanded reliability and documentation, while the emerging chatbot market moved fast. Balancing these tensions — technical depth vs. accessibility, stability vs. speed — required constant judgment calls.
We focused on making the most complex interactions feel like natural workflows. The core insight: users don't think in "entities and synonyms" — they think in "what my customer might say." So we designed the authoring experience around conversational logic rather than technical configuration. We also identified a critical gap: customers had no integrated learning path. API docs lived in one place, tutorials in another, guides in a third.
I drove the design process from low-fidelity sketches through high-fidelity prototypes, focusing on the intents and the entity and synonym creation flow — essentially the "dictionary" that trains custom AI models. The redesigned flow simplified what had been an intimidating technical process into a guided, progressive experience. I also concepted and designed an integrated learning hub that consolidated API documentation, video tutorials, and guides into a single contextual resource within the platform.
I leveraged IBM's Carbon Design System to build visually cohesive interfaces, stress-testing components, refining navigation patterns, and proposing color schemes that became integral to the Watson product family. The visual and interaction choices we shipped established a consistent identity across Watson products. Every design decision balanced IBM's enterprise rigor with the usability expectations of a modern SaaS tool.
The design decisions and product direction shaped a scalable, user-friendly platform that bridged the gap between AI complexity and business accessibility. The navigation and visual identity I helped establish became the standard across the Watson product family.
The work I contributed to Watson Conversation positioned it as a leading tool in the conversational AI space for the time. From enabling intuitive chatbot creation to fostering customer learning, the design decisions and product direction I helped shape laid the groundwork for a scalable and user-friendly platform. This project exemplifies my ability to lead cross-functional efforts, solve complex problems, and deliver designs that merge user needs with cutting-edge technology.