BCG

Overview

Utilizing GenAI to automate proccesses

The Boston Consulting Group is a management consulting firm. As a UX Design Co-op I had to opportunity to design hi-fi wireframes for a feature that allows users to configure their own agent, an autonomous system that can perform tasks. The goal of this feature is to allow users to build custom workflows to increase efficiency through automation. One of the biggest considerations was to ensure the tool could be used and understood by those with little to no technical expertise. Throughout the project, I worked with the lead design researcher, a senior designer, and the product team to understand user and business needs.

Disclaimer: Due to the sensitivity of this project, the presented deliverables have been sanitized and replaced with placeholder information to exclude all private details privy to Boston Consulting Group.

Product Overview

Features Spotlight

When I joined, I started making updates on the product based on user research. The internal product allows users to build custom workflows, share and publish them with others, use a gen AI chat bot, and more.

Chat Builder

Users can configure their build through manual setup or using an AI powered chat builder

Marketplace Preview

Builds that are published live in the marketplace where others can use and preview the build structure

Summary & Demo

At the end, users will preview their defined agent that includes metadata used and an output preview

Sharing

Users can share their builds to others in the organization as well as publish them

How Might We...

Design for 100+ data sources?

After meeting with the product team, I designed a solution to account for 100+ data sources. The previous interaction was ideal for a small amount of data sources, somewhere between 2-5. However, 100 is a more accurate reflection of the amount of sources users are dealing with.

Solving the Problem

Design Iterations

After the above feedback from product, I started to come up with different possible solutions. I met with the lead researcher and a senior designer to brainstorm and iterate on what the best approach would be.

Option 1: Table-based Interaction

The first direction I explored was providing a table of data sources that users can browse or search through.

Pros: Ability to see many data sources at once
Cons: A lot of data sources to look through if not on first page

Option 2: List View Interaction

I then explored a list option with the same functionality as the table.

Pros: No interruptions to workflow and can easily scan
Cons: Less emphasis on individual items

Option 3: Modal Interaction

Next, I did a version of the list view that would open in a modal.

Pros: Focused view on information presented
Cons: Disrupts workflow by taking users out of the page

Option 4: Search Interaction

Finally, I explored a search option where users search for data sources they want to pull from.

Pros: Quick way to find data sources
Cons: Not ideal for those who don't know which data sources they're looking for

Final Design

Building an Agent

After exploring the solutions, I combined the table-based and search interaction and designed for two use cases. The table-based interaction allows for users to browse through data sources and the search interaction allows users to go directly to the data source they want. This solution effectively caters to users who don't know what they're looking for and those that do. Below is the full workflow for adding an action to configure the build.

Reflection

Looking back, here's a couple of takeaways:

01

Push back and voice your opinions! Since I jumped into the middle of this project, I felt like I didn't have the necessary background to suggest different or new ideas. However, I learned that it's important to speak up any stage and offer new ideas, even if they don't get picked up.

02

Do the research. I was brand new to the world of Gen AI and it was super important for me to research other tools and the foundation of tools. Once I understood, I was better able to design this Gen AI tool to ensure it met the standards but also was able to stand out.

BCG GenAI

Timeline

Team

Role

Deliverables