Webex AI Assistant

Owning a design from concept through launch and all the chaos in between

Webex Control Hub AI Assistant

Who

Cisco

When

2023-2024

Role

Lead designer

Overview

Control Hub is a “Single Pane of Glass” experience for IT Admins who manage Webex and other Cisco collaboration products for their organization. We aimed to leverage the technology of large language models (LLMs) to reduce clicks, reduce steps, and reduce confusion experienced by Webex Admins who struggle to use Control Hub.

Starting with a groundbreaking product announcement at WebexOne 2023, we hit the ground running to launch a quality AI Assistant for Control Hub by June 2024.

Process

I love to build components in Figma and structure libraries in a way that enables rapid design pivots.

I was charting new territory for this project and my first step was to build out a robust set of atomic building blocks for conversational UI.

Over 9 months, these components underwent many changes, facilitated by the forethought I put into the library structure.

Vision setting

While working on the near-term product launch, I was also selected to prepare some long-term visions to present to our executive leaders. In parallel, our research team collected customer attitudes toward AI and trust.

I leveraged the research results to challenge some fundamental assumptions with our future vision. We had tried to leapfrog to an assumed future where everyone is already comfortable using AI for all of their tasks, when in reality, customers would only adopt AI products if they solve today’s problems.

I persuaded our team to tighten our AI feature priorities so that they would solve the known issues instead of simply what we were capable of building.

Conversation modeling

One of the challenges of producing a Generative AI feature is the need to balance the open-ended conversation abilities against security and privacy requirements.

After studying some fundamentals of conversation modeling, I provided our engineering team with a flowchart to guide the Assistant’s prompt suggestions. These suggestions funneled conversations into our happy paths and steered away from corners where users could get stuck or confused.

Takeaways

This was the biggest project I had worked on and a new level of responsibility and ownership. The stories and lessons I learned from this could fill more than a portfolio page.

But our successful launch can be attributed to a few key takeaways:

  • Aligning to fixed benchmarks and release dates so that design and engineering can communicate with common goals in mind

  • Becoming a fierce advocate for the customer, especially when designing for trust and the vulnerabilities of AI technology

  • Learning to be fearless in accepting new design responsibilities, more senior stakeholders, and unprecedented challenges