Sprinklr is a customer experience platform. Shipt's customer service team, called the X-Team, uses it to handle calls, chats, and emails from members, shoppers, and delivery drivers.

The company had invested heavily in Sprinklr as a replacement for their previous tool. Three months after launch it was not yielding the results expected. I was brought in to find out why.

I started by shadowing agents during live sessions. What I found was not a broken tool. It was a tool that did not know enough.

X-Team performance is measured on speed. Handle time affects their metrics and their pay. So agents had built their own workarounds. Admin, Shipt's internal tool, was always open in a separate tab because almost everything they needed during a call lived there, not in Sprinklr.

When I presented to directors and VPs, I showed them a matrix mapping tool adoption against usage frequency. The goal state was clear: all agents using Sprinklr, all the time. But based on what I observed during shadow sessions, I was able to plot where we actually were. Most agents were using it partially, working around it for anything that required speed. The matrix made the gap visible in a way that a list of problems could not.

The gap came down to a few things, some of which they were already aware of and some they weren't.

Working inside Sprinklr meant we could not use our full design system. Their components, their visual language. We had access to a specific slice of the codebase and not much more. So instead of leading with UI, we led with information. What we surfaced, how we prioritized it, and in what order became the design. That constraint pushed us to be more intentional than we might have been otherwise.

This shifted our entire design focus. We were not designing UI. We were designing UX. Readability, cognitive load, ease of use, and accessibility became the primary design decisions because they were the ones we actually owned.

The solution was a widget embedded in the active case view. When a call or chat came in, Sprinklr would detect who the caller was. If there were multiple account matches, the widget listed them so the agent could verify quickly with the caller. Faster than searching admin and navigating a results page.

The discovery work for this project ended up informing a roadmap that was not ours.

The wrap up code problem led to an idea I brought to engineering. The transcript and chat data already existed. We could use behavioral pattern detection to pre-select the most likely wrap up codes, leaving agents to confirm or edit rather than fill from scratch. This was our first experimentation with AI at Shipt, before the company had a formal AI feature strategy. It was born from a simple observation. The data was already there. We just needed to use it.

The canned response problem, the timestamp problem, and several others I surfaced became officially roadmapped features for the X-Team product. A year and a half later that team still reaches out to reference the original research when they are building new features.

The widget opened a door. We began planning personalized widgets for each agent type. Members, shoppers, and last mile delivery drivers all have different needs. A shopper calling about a denied transaction should land directly on their payment page, not their general profile.

The immediate follow-on project was chatbot and IVR parity. I designed the chatbot experience and the IVR team followed our interaction model to create consistency between text and voice. The goal was to resolve more cases before they reached a human agent.

The Sprinklr investment started showing returns. X-Team directors and leaders cited cost reduction as a direct goal met. The research I did in those first three months became the foundation for two teams' roadmaps, not one.

The tool was not working as expected and we found a way to optimize it. But more than that we set a blueprint for what the integration could become, with a clear vision and actionable next steps already in motion.

Designing inside someone else's platform forced a clarity that is easy to lose when you own everything. We could not hide behind visual decisions. Every choice had to earn its place through usefulness alone.