At Microsoft, I worked on Cora, an enterprise chatbot and sales assistant designed to support business users through natural human-to-machine interaction.

This work happened years before the current generative AI wave. Conversational interfaces were still emerging, and many of the patterns now considered familiar had not yet become standard practice. The work required asking foundational questions about how people interact with intelligent systems: what the system should know, how it should respond, what personality it should express, and how users could understand what it was capable of doing.

My role centered on designing Cora’s personality, interaction model, voice and GUI response patterns, high-fidelity prototypes, and supporting design specifications.

Summary

Role: Senior UX / Product Design
Focus: Enterprise chatbot personality, voice interaction, GUI prompting, sales assistant workflows
Environment: Microsoft, enterprise AI, conversational interfaces, LUIS, Teams
Core contribution: Designed Cora’s personality, interaction behavior, response patterns, and multimodal experience across voice and visual interfaces.

The Opportunity

Cora was not simply a chatbot interface. It was an early attempt to make an enterprise system feel more natural, responsive, and useful to the people who depended on it.

The opportunity was to create a sales assistant that could support users across modalities, including graphical interfaces and voice-only interaction. This required designing not only what users saw, but what the system said, how it behaved, how it clarified intent, and how it maintained trust.

The challenge was to make artificial intelligence practical, understandable, and useful inside a real business context.

What I Worked On

My work included:

• Defining Cora’s personality and behavioral attributes

• Designing her as prepared, trustworthy, intelligent, and personable

• Creating speech patterns based on platform logic and her role as a sales assistant

• Designing interaction patterns across GUI and voice-only modalities

• Supporting hands-free workflows through voice-enabled responses

• Creating high-fidelity prototypes for key scenarios

• Translating whiteboard concepts into research-ready prototypes and agile user stories

• Creating detailed specifications for engineering and sprint planning

• Designing UI, UX, illustration, and supporting marketing material

The work required close collaboration across business analysis, user research, engineering, and platform architecture.

Finding the Frontier

One of the things that helped me unlock the Cora work was self-directed learning outside the normal project path. While working at Microsoft, I spent additional time learning from the emerging voice-interface community, including Amazon Alexa open forums and free onsite training opportunities. I was not formally part of the Alexa team, but I wanted to understand how the strongest voice teams were thinking about conversational patterns, intent structures, user prompts, and voice-only interaction.

That time at the frontier changed how I approached the work. It helped me think beyond screens and apply voice-first thinking to an enterprise assistant that needed to support both visual prompting and hands-free interaction. The work became less about designing a chatbot interface and more about understanding intent, clarification, trust, personality, and how intelligent assistants should behave when language becomes the interface.

I brought those learnings back into Cora, where they helped shape the interaction model, voice and GUI response patterns, and personality framework. That experience became part of a larger thread in my work: exploring how people and increasingly intelligent systems communicate in ways that are useful, understandable, and grounded in human needs.

Human-Machine Communication

One of the most important aspects of the work was designing how Cora communicated.

Personality in an enterprise assistant is not decoration. It shapes trust, comprehension, confidence, and adoption. A system that sounds too robotic can feel brittle. A system that sounds too casual can feel unreliable. The right balance depends on the job the system is doing and the context in which people use it.

For Cora, that meant creating a personality that supported her business role. She needed to be useful without being intrusive, intelligent without pretending to be human, and personable without undermining trust.

The work also required designing across modalities. A GUI can provide visual scaffolding, reminders, and fallback paths. A voice-only experience has fewer visible affordances and depends more heavily on clarity, pacing, confirmation, and recovery. Designing Cora meant considering both modes together, so the assistant could support users whether they were reading, listening, speaking, clicking, or working hands-free.

One of the most important aspects of the work was designing how Cora communicated.

Personality in an enterprise assistant is not decoration. It shapes trust, comprehension, confidence, and adoption. A system that sounds too robotic can feel brittle. A system that sounds too casual can feel unreliable. The right balance depends on the job the system is doing and the context in which people use it.

For Cora, that meant creating a personality that supported her business role. She needed to be useful without being intrusive, intelligent without pretending to be human, and personable without undermining trust.

The work also required designing across modalities. A GUI can provide visual scaffolding, reminders, and fallback paths. A voice-only experience has fewer visible affordances and depends more heavily on clarity, pacing, confirmation, and recovery. Designing Cora meant considering both modes together, so the assistant could support users whether they were reading, listening, speaking, clicking, or working hands-free.

Rapid Ideation One of the most important aspects of the work was designing how Cora communicated.

Personality in an enterprise assistant is not decoration. It shapes trust, comprehension, confidence, and adoption. A system that sounds too robotic can feel brittle. A system that sounds too casual can feel unreliable. The right balance depends on the job the system is doing and the context in which people use it.

For Cora, that meant creating a personality that supported her business role. She needed to be useful without being intrusive, intelligent without pretending to be human, and personable without undermining trust.

The work also required designing across modalities. A GUI can provide visual scaffolding, reminders, and fallback paths. A voice-only experience has fewer visible affordances and depends more heavily on clarity, pacing, confirmation, and recovery. Designing Cora meant considering both modes together, so the assistant could support users whether they were reading, listening, speaking, clicking, or working hands-free.