Finding new use cases for AI-powered HR assistant

overview

I led part of the discovery and research for Sage, Kroger’s AI-powered HR assistant.

Initially built for HR Specialists, Sage’s scope was being expanded to assist in-store associates with their daily HR tasks.

I was charged with figuring out what new use cases for Sage would be most relevant for in-store associates.

Intro 

We conducted a large-scale survey 5 months after launch of Sage, an associate-facing AI chatbot to assist with HR functions. The purpose was to see if the assumptions made about user needs at the beginning of the Sage project are matching up with how associates are using Sage today.

Kroger / 2025 / My Role: Product Designer and Researcher

Challenge

Five months after the launch of Sage, the company’s associate-facing AI chatbot, we were interested to see if Sage was meeting the needs of associates as intended.

I was officially part of the Product Design team, but 90% of my role was as a researcher, leading the survey development, analysis, and aligning the scope and results with company goals.

Company goals I helped support:

• Increasing associate retention

• Reducing time from application to hiring

• Create a smoother onboarding experience

Discovery & Research Questions

We aligned with the stakeholders and team on our purpose, and started developing the survey.

Our research questions were:

Personas

The personas we were focusing on fell into two categories, Schedule Editors, who can edit work schedules, and Hourly Associates, who are regular in-store workers with no schedule editing power.

Survey Development

The survey writing process for this project looked like this:

Following this survey-writing process ensures that we are aligned with our research goals, avoid leading questions, and ensure that the survey is very clear, as it is a self-standing item.

response criteria

We asked associates to rate various functions by Time (time on task), Importance, and Satisfaction:

Open-Ended Questions

Anticipating the volume of responses, we kept open-ended questions to a minimum, but opened it up for feedback that we may have not anticipated.

Synthesis & Findings

Received 19,000 responses - an amazing turnout!

• This is a representative sample of the HR specialist / in-store associate populations.

A few key problems:

• There are several issues with the clock-in system, and nearly all associates have had direct problems with it.

• Managing time off is difficult, for both those requesting time off and those approving them.

These results validated the assumptions we made about associate needs at the beginning of the Sage project.

comparing survey to usage data

• In order to validate the survey data, we cross-referenced the results with usage data from the Sage AI chatbot. We had about 3-4 months of data since launch, and they were categorized based on what the associates would ask.

The two datasets agreed with each other!

• Top 4 issues indicated in the survey agreed with the highest request topics within Sage usage data, for both persona groups.

The Outcome

This study helped our team in a few ways:

• Helped validate that our assumptions about associate needs were correct

• The large number of responses helped our team get more buy-in from upper management stakeholders and we were able to acquire additional developer hours towards the project.

• Helped the Sage project become a high visibility project

Overall, the survey was a success — we were cleared for another survey 6 months later, and to increase the scope to include additional personas. With the additional developer resources we secured, we could push forward with more features for Sage.