Transforming StoreOps through AI

Designing an internal AI assistant under extreme ambiguity

Read case study

Learn about me

$700K

gains in annualized labor allocation

90.9%

satisfaction rate on the first week

587

stores piloted

5,000+

AI conversations

Transforming store operations through AI, delivering $700K in annualized labor gains

Role

Sr. Product Designer

Industry

Retail

Length

5 months

Team

Aboutthisproject

This initiative focused on empowering partners by giving them faster access to reliable, Petco-backed information through an AI-powered assistant. On average, only two associates are available per store at a time, often leading to delayed answers and frustrated customers. By leveraging AI, the goal was to reduce the time associates spend finding data, increase customer trust, and improve overall operational efficiency.

Myrole

Senior Product Designer (end-to-end ownership). I led discovery with PMs and stakeholders, facilitated workshops, synthesized insights, defined flows, prototyped, and designed the end-to-end experience. Worked within a cross-functional team of 10+ people, including engineers, PMs, stakeholders

User Research

Analysis

Cross-functional collaboration

Workshop lead

User flow

mapping

Wireframing

Understandingthestartingpoint

Constraints & reality

•No time for foundational research. •No baseline data. •No agreed definition of “done”. •Many ideas, most out of scope. •A small research effort already completed, but with conclusions already locked in.

A project that started with almost no structure

Business problem

Constraints & reality

The direction (AI assistant) was decided early due to strong political pressure to move along with AI - The value was not.

Understandingthestartingpoint

Defining a minimal, opinionated North Star

Instead of trying to solve everything, I deliberately scoped v1 to: • Answer the most common partner questions. • Reduce time-to-answer. • Require the least cognitive effort possible. I intentionally reduced content and functionality to the minimum viable experience, accepting uncertainty in exchange for speed and clarity.

Design

Designing without perfect research

Turning metrics into research

Reprioritizing based on evidence, not opinions

The direction (AI assistant) was decided early due to strong political pressure to move along with AI - The value was not.

Key design decisions (ownership)

Trade-offs & risks

What we gained

High adoption and satisfaction

Clear evidence of value through labor reallocation
Buy-in from teams beyond the original stakeholders

What we accepted

Designing the company’s first AI tool with minimal precedent

Shipping with incomplete certainty
Letting metrics — not upfront research — guide discovery.

What we didn’t get

A voice-first experience, which I advocated for to support on-the-go usage, but was constrained by technical scope

Impact