Ada is the AI customer service chatbot used by 350+ businesses, including Square and Wealth Simple.

Over the last 5 years I’ve worked on numerous projects to improve both the chatter and client-facing experience.

Chat redesign

In 2019, Ada’s web chat was looking outdated, clunky, and lacked brand visibility. Clients were adding custom CSS to make the chat adhere to their brand, which prevented code migrations.

As the sole designer, I led the redesign over a period of 4 months.

I designed the new chat with 4 principles in mind — brandability, seamlessness, accessibility, and a fresh look and feel.

The new chat was much more compact in size. We also added a suite of customization options that were top requests from clients such as header styles, font size scaling, proactive messaging bubbles, and light/dark mode.

Redesigning the chat meant thinking through every aspect of the user journey, such as agent handoff and post-chat surveys. We built a new design system that was modular and flexible.

Some challenges include designing components to be adaptable for 3rd party widget, automatically defaulting colors for accessibility compliance (WCAG 2.1), and ensuring the chat scales properly across screen sizes and devices.

Along with the chat, I designed the new customization settings to be simple to configure and improved findability.

The entire chat redesign allows us to:

  • Increase deals and contract sizes for customers with advanced branding needs

  • Migrate customers off of custom code

  • Reduce support tickets for bugs and customizations requests

Generative AI features

With the arrival of chatGPT in 2022, we pivoted from a drag-and-drop content builder to plug and play. Businesses can simply add their knowledge base and the “AI Agent” will use it to generate answers. I designed the entire knowledge ingestion and in-product article creation flow.

I worked on “Rules” which allows businesses to conditionally define when their AI Agent can use a specific set of knowledge articles. Rules can be grouped and nested for complex logic.

Users needed a window into the AI’s decision making so that they can understand how content is being used and make improvements.

I explored different ways of displaying the AI’s reasoning in the conversations view, user tested with customers, ultimately landing on displaying it inline for better scannability.

Exploration 1 - display reasoning steps in sidebar

Exploration 2 - display reasoning steps inline

Channel and integration expansion

I led design on our channel expansion to support conversations over SMS, Instagram, WhatsApp, FB messenger, and more. No matter the channel, we made sure that different message types will gracefully default according to the platform’s limitations.

In 2022, we added voice support. I designed the early conversational UI for voice.

In 2024, I designed the self-serve integrations experience which allows users to discover and install integrations without the help of Ada staff. This experience also unified the dashboard experience and allows 3rd party developers to easily build on our platform.