ChatKit

Role
Product Design Lead

TImeline
2023 - current

Team
Product Manager
Developers
Solutions Architect
Accessibility Team
Designers

Project Goals
Currently, creating a chatbot for small businesses or internal use at Apple involves partners requesting a chatbot from our team. They manually build the bot using dialogue builders through our application, Studio, and work separately with the ML team to develop tailored intents and models. Our goal is to simplify and automate the chatbot creation process, enabling anyone to create a chatbot—whether for business or demonstration purposes—more autonomously.

Impact
Small businesses associated with Apple can use this product to create their own chatbots for answering customer inquiries.

Internally, various business units within Apple often seek chatbot solutions for their products. With ChatKit, we provide these businesses the autonomy to design and manage their own chatbots, ensuring alignment with their specific control requirements and business standards.

Problem

Without generative AI, this is how complex the platform got for building a simple conversation solution.
The users would get lost in this infinite canvas and not know how to manage conversations.

It’s a long process to create a simple chatbot

Needs

Customer Needs

  • Easy to use product

  • An organized product

  • Accessible

  • Intuitiveness

  • Simplicity

Business Needs

  • Cost savings for team resourcing

  • Our team no longer needs to handle all the tedious aspects of chatbot building.

Opportunity

How might we use generative AI to simplify conversation building platforms for inclusive, intuitive experiences?

Brand Study

Goals
To gain deeper insights into which brand name resonates most effectively with our product and organization, we conducted two surveys at Cafe Macs. The majority of the participants were from the Marcom team.

Study 1
“ChatKit" was perceived as having the feel of an Apple product and was associated with messaging, communication, AI, and API tools. In contrast, "Ckit" was considered less descriptive, with participants unclear about its function.

Study 2
Each brand name received unique feedback, yet both "Conversation Kit" and "ChatKit" were generally perceived as relating to chatbots, messaging, conversations, and AI. The key distinction was that "ChatKit" conveyed a more informal, "Apple-like" brand feel. Additionally, "Studio AI" was frequently associated with creative studio tools.

Iterations and playground

Initial Iteration

Feedback
The team supports the new approach of integrating the existing application (Customize Flow) with the generative model. Additional feedback suggests refining it to better align with Apple's brand and style.

Next Iteration of Onboarding

These are the serious of questions the ML team need to create a chatbot. Starting with the Bot characteristics such as the title and image, bot platform, country and language, volume of users, the model selection, behavior instructions, document uploads, and any APIs

Feedback
The team appreciates the color combination but is still working to ensure it aligns with Apple’s standards while incorporating our team’s branding and identity.

Current Designs

🎉 My Takeaways

I learned to lead product designers and make design decisions based on not only user needs but also product needs. I’ve learned to compromise with product managers and engineers to find the common ground for development.