Amdocs AI copilot

The client

Amdocs is a global leader in software and services for communications, media, and financial services providers. Amdocs serves most of the world's largest telecom operators, including AT&T, Vodafone, Verizon, and Telefónica.

It's also where I’m proud to work.

Overview

Following the 2023 boom in generative AI, Amdocs decided to enhance its products with a generative AI copilot named amAIz, the world's first telco-native AI solution.

I was tasked with the UX design for this generic solution, which is now being implemented across multiple Amdocs solutions. Two months after the design was completed, this solution was presented at the world-leading Telco conference, MWC.

My role

As a UX researcher and designer, I was tasked with defining the entire solution framework and behavior, delivering high-fidelity wireframes and documentation for development. I partnered with a talented UX/UI designer who translated my wireframes into the final UI visual design.

My role included:

Collaborating with product management to define the scope
Researching industry standards and best practices
Defining generic behavior to support complex use cases on multiple platforms
Creating UX concepts and detailed designs (wireframes and full documentation)
Handing over to multiple design teams who worked on specific use cases

Design highlights

Flexible framework

Supporting multiple separate apps and different contexts of use, we defined four basic panel states:

Responsive drawer - Pushes other content (enabled only on responsive applications that can support the copilot’s minimum width).
Docked floating panel / resizable floating panel - Docked to the right or left side of the page, or floating above any part of the page, while hiding other content.
Full page
Mobile full page

Sensitive business information has been redacted

Responsive drawer

Floating panel

Full page

Mobile

Actions

Supporting highly complex and technical apps, we needed to ensure that actions on the Copilot were both precise and seamless for our users.

To achieve this, we defined some basic rules:

Actions affecting content outside the Copilot will always be visually differentiated.
Actions based on user natural language requests will be interpreted, presented, and confirmed by the user before execution to prevent unwanted outcomes.
An undo option will always be available, allowing users to easily roll back in case of an unintended outcome.