Overview
Objectives
Make simple customer service problems effortless to resolve and complex problems easier to navigate.
0 -> 1 design leveraging AI and guided workflows to automate routine resolutions, reduce cognitive load, and provide Online Customer Service Associates with contextual guidance throughout the customer interaction.
Thoughts & Feelings
I was excited by the opportunity to explore how AI could support associates without replacing the human side of customer service. At the same time, I knew the challenge went beyond designing an interface—we needed to understand complex service workflows and determine where automation could genuinely reduce friction.
Going into the project, my focus was on creating an experience that felt supportive rather than prescriptive: giving associates the right guidance at the right moment while still allowing them to use their own judgment when situations became more complex.
My Role & Responsibilites
As Lead Designer, I owned the end-to-end UX process—from discovery and workflow definition through prototyping, usability testing, final UI, and engineering handoff.
Tools
Figma, Confluence,
Sprint
6 Months
Team
1 Product Manager · 10 Engineers · 4 Business Stakeholders
About HeyHomer: Guided Workflows
HeyHomer began as an AI-powered tool helping store associates quickly access product information, inventory, SKUs, fulfillment options, and vendor knowledge.
Guided Workflows extended that intelligence into Online Customer Service, bringing customer communication, order management, institutional knowledge, and resolution guidance into a unified experience.
Rather than simply answering questions, HeyHomer could help associates understand the customer’s situation, determine what needed to happen next, and guide them toward an appropriate resolution.
The Problem
Online Customer Service Associates were expected to resolve a wide range of customer issues, but they had no centralized knowledge base or standardized resolution process.
Much of the knowledge required to do the job was tribal—learned during initial training, accumulated through experience, or gathered by asking more experienced associates for help in team support chats.
At the same time, associates had to move between multiple tools while actively assisting customers:
HOME — Order Management
Used to locate orders and SKUs, process returns, order replacement parts, and issue customer accommodations such as gift cards.
External Customer Communication Platform
A separate system used to manage incoming customer calls and online chat conversations.
Team Support & Tribal Knowledge
When associates didn't know how to resolve an issue, they often relied on coworkers, managers, or knowledge retained from training.
This created a service model that depended heavily on experience and institutional knowledge rather than the tools themselves.
What if the expertise of a senior associate could be built directly into the experience?
The Business Challenge
Expertise was valuable—but difficult to scale.
The service model relied heavily on training, tenure, and institutional knowledge. As support volume grew, so did the cost and complexity of maintaining consistent service.
Long ramp-up times — New associates needed extensive training before they could confidently handle a broad range of customer issues.
Inconsistent resolutions — Without standardized guidance, the path to resolving the same issue could vary between associates.
Higher cognitive load — Associates had to remember policies, navigate multiple tools, communicate with customers, and determine the appropriate resolution simultaneously.
Limited escalation paths — Complex situations were often routed through managers rather than directly to the teams best equipped to resolve them.
Senior associate dependency — Experienced associates became an informal knowledge base for newer team members, pulling them away from higher-value and more complex customer problems.
Operational cost — Training, longer interactions, unnecessary escalations, and reliance on highly experienced associates increased the cost of servicing customers.
The Opportunity
What if the expertise of a senior associate could be built directly into the experience?
Guided Workflows aimed to consolidate customer communication, order management, institutional knowledge, and resolution workflows into a single AI-assisted experience.
Instead of requiring associates to memorize how to resolve hundreds of potential scenarios, the system could understand the customer issue, surface relevant information, guide the associate through the appropriate resolution, and automate routine actions when possible.
The goal wasn't simply to make experienced associates faster.
It was to help entry-level associates perform more like experienced ones.
Discovery & Research
Understanding how experts make decisions.
I worked with associates, managers, SMEs, and business stakeholders to understand not only what associates did, but the knowledge and decisions behind each resolution.
Through contextual inquiry, interviews, workflow mapping, and collaborative workshops, I decomposed common service scenarios into their underlying questions, decision points, business rules, actions, and escalation paths.
Defining the System
Turning complex service scenarios into a repeatable framework.
Discovery showed that although customer issues varied, most resolutions followed the same underlying pattern. I used that insight to define a shared workflow architecture that could support multiple scenarios without redesigning the experience each time.
Designing the Experience
Turning the system into an experience associates could use in real time.
With the workflow architecture established, I focused on translating complex business logic into an experience that felt simple, predictable, and supportive during live customer interactions.Instead of exposing the complexity behind each resolution, Guided Workflows progressively revealed only the information and actions associates needed at each stage.
Instead of exposing the complexity behind each resolution, Guided Workflows progressively revealed only the information and actions associates needed at each stage.
One workspace, not another tool
Bring the customer conversation and resolution workflow together.
Associates could manage their queue, communicate with the customer, access relevant context, and work toward a resolution without constantly switching between systems.
Design goal: Reduce context switching and keep attention on the customer.
Progressive guidance
Turn complex processes into manageable steps.
The workflow progressively guides associates through: Verify → Identify → Understand → Resolve → Confirm
Each step builds on information gathered previously, reducing the amount an associate needs to remember or determine independently.
Design goal: Make complex workflows approachable without overwhelming the associate.
AI at the moment of need
Support the conversation without taking it over.
AI-generated prompts provide contextual guidance based on where the associate is in the workflow—helping them ask the right questions and move the interaction forward.
The associate remains responsible for the customer relationship and final decision.
Design goal: Augment human judgment rather than replace it.
Validate & Iterate
Testing the system, not just the interface
I validated Guided Workflows with Customer Service Associates using realistic support scenarios. Rather than testing individual screens in isolation, I focused on whether associates could move confidently from customer intent → issue identification → resolution while maintaining the conversation.
Make progress visable
What we learned
Associates needed a stronger sense of where they were in the resolution process and what remained.
How I responded
I moved from a linear checklist toward clearly defined workflow stages with active, completed, and upcoming states.
Design principle: Always make the next step clear.
Make AI contextual
What we learned
Generic AI assistance wasn't enough. Guidance became significantly more useful when it reflected the customer's issue, conversation, and current workflow stage.
How I responded
I designed Suggested Prompts to respond dynamically to conversation context and workflow state—helping associates ask the right questions without scripting the entire interaction.
Design principle: AI should reduce thinking required, not remove human judgment.
Design for exceptions
What we learned
Real customer service doesn't always follow the happy path. Associates needed a way forward when the workflow couldn't confidently determine the appropriate resolution.
How I responded
I designed explicit escape hatches into the system, allowing associates to consult an expert or escalate directly to the appropriate partner without abandoning the customer interaction.
Design principle: Automation should know when to get out of the way.
We weren't designing the perfect happy path. We were designing a system associates could trust when the path wasn't clear.
Impact & Strategic Value
Designing for impact beyond the individual interaction
Guided Workflows was designed to create value at multiple levels of the service organization. By embedding institutional knowledge and business logic directly into the associate experience, we could reduce dependence on tenure while creating more consistent and efficient paths to resolution.
The impact extended beyond associate productivity. Standardizing how knowledge was surfaced, decisions were made, and exceptions were routed created a foundation for scaling expertise across customer service.
Measuring impact
We compared baseline workflows with Guided Workflows across representative customer-service scenarios. Resolution time, consistency, escalation behavior, and help-seeking were evaluated through validation and pilot activity, while longer-term measures such as time-to-proficiency were modeled against existing training benchmarks.
Pilot Validation
We compared the existing support experience with Guided Workflows across representative customer-service scenarios. The pilot focused on whether contextual guidance could help associates resolve cases faster, follow more consistent resolution paths, and rely less on escalation or outside support.
Turning experience improvements into operational leverage
At enterprise scale, improvements measured in minutes and percentage points compound across customer interactions.
Reduced handling time, fewer escalations, and faster associate proficiency created the potential to increase service capacity without scaling operating costs at the same rate.
My Contribution
The Strategic Shift
I connected associate pain points, operational constraints, and business goals into a scalable workflow model—then partnered across Product, Engineering, and Operations to translate that strategy into an experience capable of supporting multiple resolution types.
The Strategic Shift
The most important outcome wasn't a single workflow or efficiency gain. It was changing where expertise lived within the service organization.
The bigger outcome wasn't simply faster associates—it was a more scalable service model.
Final Experience
Reflections & Learnings
What I learned
Guided Workflows reinforced that the most valuable application of AI wasn't replacing associate decision-making—it was making organizational knowledge available at the exact moment a decision needed to be made.
My role evolved beyond designing an interface. I worked across associates, product, engineering, and business stakeholders to define the underlying service model, translate complex operational rules into reusable patterns, and establish a framework that could scale beyond the initial workflows.
Takeaways
Design the system, not just the screen.
The breakthrough came from identifying reusable patterns beneath seemingly different customer problems.
AI needs boundaries.
Automation became more valuable when associates understood what the system knew, what it recommended, and when human judgment should take over.
Enterprise UX is organizational design.
Changing the associate experience also changed how knowledge, expertise, and decision-making could move through the organization.
AI at THD
One piece of the puzzle
Guided Workflows was part of a company wide initiative empowering our customers and associates with AI. Check out how this project worked alongside others,
.png)
