G-P AI
COMPANY
G-P
ROLE
UI Lead
TIMEFRAME
Aug 23 - Present
G-P AI
COMPANY
G-P
ROLE
UI Lead
TIMEFRAME
Aug 23 - Present
G-P AI
COMPANY
G-P
ROLE
UI Lead
TIMEFRAME
Aug 23 - Present
CHALLENGE 01
Finding design solutions that work for extremely diverse use cases across functions
CHALLENGE 01
Finding design solutions that work for extremely diverse use cases across functions
CHALLENGE 01
Finding design solutions that work for extremely diverse use cases across functions
CHALLENGE 02
Designing an approach that thinks big but is considerate of the technology available
CHALLENGE 02
Designing an approach that thinks big but is considerate of the technology available
CHALLENGE 02
Designing an approach that thinks big but is considerate of the technology available
CHALLENGE 03
Designing for and managing dozens of excited stakeholders across functions
CHALLENGE 03
Designing for and managing dozens of excited stakeholders across functions
CHALLENGE 03
Designing for and managing dozens of excited stakeholders across functions

Brief

GIA is an ambitious project initiated by G-P to harness the potential of artificial intelligence in enhancing the digital experience for users across our platform. The goal was to integrate AI technologies in a way that feels natural and supportive, rather than overwhelming, enriching the platform's capabilities and assisting both users and internal teams.

GIA is an ambitious project initiated by G-P to harness the potential of artificial intelligence in enhancing the digital experience for users across our platform. The goal was to integrate AI technologies in a way that feels natural and supportive, rather than overwhelming, enriching the platform's capabilities and assisting both users and internal teams.

Team

GIA is supported by a cross-functional team, bringing together expertise from engineering, data analysis, operations, and customer support, among other departments.

This collaborative effort unites front-end developers responsible for the user interface, back-end developers handling data architecture, product managers guiding the project's direction and of course myself as the lead designer on the project, designing the user interface.

GIA is supported by a cross-functional team, bringing together expertise from engineering, data analysis, operations, and customer support, among other departments.

This collaborative effort unites front-end developers responsible for the user interface, back-end developers handling data architecture, product managers guiding the project's direction and of course myself as the lead designer on the project, designing the user interface.

USER 01
Current customers of G-P who are using the platform and need assistance day-to-day that doesn't rely on human support.
USER 01
Current customers of G-P who are using the platform and need assistance day-to-day that doesn't rely on human support.
USER 01
Current customers of G-P who are using the platform and need assistance day-to-day that doesn't rely on human support.
USER 02
Professional users of G-P who only intermittently use G-P, frequently have complex questions that can't be easily answered.
USER 02
Professional users of G-P who only intermittently use G-P, frequently have complex questions that can't be easily answered.
USER 02
Professional users of G-P who only intermittently use G-P, frequently have complex questions that can't be easily answered.
USER 03
Sales prospects, interested in G-P's products and services and need fast answers to complex questions without waiting for a call
USER 03
Sales prospects, interested in G-P's products and services and need fast answers to complex questions without waiting for a call
USER 03
Sales prospects, interested in G-P's products and services and need fast answers to complex questions without waiting for a call
USER 04
Internal G-P teams who would use our AI solutions to find answers to complex questions themselves, cutting down on internal tickets.
USER 04
Internal G-P teams who would use our AI solutions to find answers to complex questions themselves, cutting down on internal tickets.
USER 04
Internal G-P teams who would use our AI solutions to find answers to complex questions themselves, cutting down on internal tickets.

Approach

Our approach to designing GIA was therefore very collaborative.

We aligned with stakeholders across departments, from product management to data science, to pinpoint where AI could add the most value. Our product strategy focused on leveraging G-P's rich internal data to inform our AI features, ensuring they would be both practical and beneficial while using the enthusiasm around AI to take a very blue sky approach to solving the basic problems.

We felt strongly that GIA would be a chance to explore the future of the platform and that thinking big and bold would be the right place to start. This was a rare project where compromise wasn’t really on anyone’s minds. Every team involved wanted to push boundaries.

Research

User research was conducted through A/B testing as well as moderated and unmoderated interviews and user tests to gauge overall reception and (more importantly with new technology) understanding of our proposed solutions.

This involved both potential and existing users to provide a well-rounded understanding. While this research is ongoing, it has provided useful insights. One of which being that while it may be following a trend, mentioning “AI” inside of chat bot, command bars, etc. would be important to help users to understand how to interact with each feature.

We commissioned Gartner to conduct additional research and their findings matched ours: if users don’t see “AI” emblazoned on a chat feature or search they will assume limited pre-AI functionality. We also found in testing that due to the newness of AI technology, users generally didn’t understand yet how to interact with AI. As a result, we decided to limit chat interactions and not fully integrate AI into guided flows and form submits. Instead, AI would be there as an assistant rather than a data gatherer.

Solutions

This iterative design and validation process led us to a refined set of AI features.

We decided to start with AI chat and a command bar. AI chat would be for longer conversations (like a personal assistant) that had a memory of the users’ previous interactions and information. The command bar was focused on fast knowledge assistance and AI-powered search.


We also launched an AI-assisted feature integration into our Salary Insights product which would allow users to find what their salary should be in a specific city.


This was a fairly simple first-pass integration but we felt that the impact similar features would have on the platform could be huge so we included it in our first round of priorities.

GIA is integrated throughout G-P. Here it's a floating action button that reveals GIA chat.

GIA is integrated throughout G-P. Here it's a floating action button that reveals GIA chat.

GIA exists as a command bar, chat application and is directly integrated into features throughout the platform.

GIA exists as a command bar, chat application and is directly integrated into features throughout the platform.

10+

Teams involved in bringing GIA to life: collating and providing data, assisting with research and finding use cases for GIA within their products.

150+

Internal users of the first Slack-based iteration of GIA before launch.

Here GIA is docked into our guided overlay flows, providing context and answering questions during onboarding and request processes.

Here GIA is docked into our guided overlay flows, providing context and answering questions during onboarding and request processes.

Chat Design

Chat Design

We designed GIA chat to be as flexible as possible so that it could handle an endless amount of potential user interactions, AI responses, and even hallucinations easily.

We designed GIA chat to be as flexible as possible so that it could handle an endless amount of potential user interactions, AI responses, and even hallucinations easily.

The GIA chat application on first launch. GIA is context-aware and welcomes users with a message that understands what they're currently doing in the platform..

The GIA chat application on first launch. GIA is context-aware and welcomes users with a message that understands what they're currently doing in the platform..

Handover to human customer support is directly integrated into the GIA app using Intercom's API. This ensures that GIA is the one-stop shop for all knowledge and customer support.

Handover to human customer support is directly integrated into the GIA app using Intercom's API. This ensures that GIA is the one-stop shop for all knowledge and customer support.

Chat history is also directly built into the app so users can find their previous chats with GIA easily. Chats are automatically named based on context and sorted by date.

Chat history is also directly built into the app so users can find their previous chats with GIA easily. Chats are automatically named based on context and sorted by date.

To ensure we steadily improve GIA's responses to customers, feedback options are provided on every GIA response. When tapped, a floating feedback dialog appears.

To ensure we steadily improve GIA's responses to customers, feedback options are provided on every GIA response. When tapped, a floating feedback dialog appears.

CONTACT

Email Me

CONTACT

Book Time

LINKEDIN
DRIBBBLE
MEDIUM
CONTACT

Email Me

CONTACT

Book Time

LINKEDIN
DRIBBBLE
MEDIUM
CONTACT

Email Me

CONTACT

Call Me

LINKEDIN
DRIBBBLE
MEDIUM