Monica Alexander — UX Designer

Introduction

Designing a reusable interaction model for voice, touch, and screen based Alexa experiences.

Context

I designed an object based model that defines what customers interact with, what actions they can take, and how those interactions behave consistently across the family of Alexa devices.

The problem

Alexa's customer experience had historically been shaped by device specific teams making device specific flows. These business fractures showed in the interface because actions were behaving differently across viewport profiles and during a noun forage, we were seeing different UX writing applications for capabilities across devices.

Customers may interact with the same capability through voice, touch, or screen UI but the structure, naming, and actions were not aligned. This created fragmentation for customers and an internal inefficiency for experience makers across devices.

Principles

• Objects must behave predictably across contexts.
• A single clear way to complete actions.
• Consistency over novelty.
• Structure supports learning and memory over time.
• Experiences adapt to context while maintaining familiarity.

01 · Context

Context

I designed an object based model that defines what customers interact with, what actions they can take, and how those interactions behave consistently across the family of Alexa devices.

02 · The problem

The problem

Alexa's customer experience had historically been shaped by device specific teams making device specific flows. These business fractures showed in the interface because actions were behaving differently across viewport profiles and during a noun forage, we were seeing different UX writing applications for capabilities across devices.

Customers may interact with the same capability through voice, touch, or screen UI but the structure, naming, and actions were not aligned. This created fragmentation for customers and an internal inefficiency for experience makers across devices.

03 · Key insight

Key insight

Customers don't think in screens and internal systems. Customers think in things. Customers think in lights, rooms, routines, music, timers, calculators, reminders, and settings. By modeling these known words as objects with shared actions and states, the experience could become more predictable across each existing device and support upcoming launches.

04 · Key design decision

Key design decision

I chose to define objects based on how customers naturally refer to things such as lights, rooms, and routines, rather than how our business is structured. This reduces cognitive load and made interactions more predictable across devices.

05 · Goal

Goal

Create a reusable model that defines the Alexa experiences around objects, actions, and states so that experience makers can design more consistent experiences across devices and inputs.

06 · Approach

Approach

I mapped repeated nouns across Alexa experiences to identify the core objects that customers interact with. From there, I defined how each object behaves. I outlined the attributes, what actions customers can take, and what states the system should return. This helped move our design conversations away from isolated screens and towards reusable interaction patterns.

07 · System model

System model

Object > Action > State

Each object was defined in three parts. An Object is the thing customers interact with. Actions are what customers can do to an object. States are how the object changes or responds.

An example is a Smart Home Light. A customer can take action by turning on, turning off, or dimming the light. The state of the light would change on, off, or even dimmed. The customer can do this through various inputs such as voice, touch, or on their remote.

08 · Object cards

Object cards

To make the model usable, each object needs a clear definition. Object cards documented the object's attributes, available actions, possible states, and supported input methods.

09 · Cross device behavior

Cross device behavior

The model clarified how the same object should behave differently across different inputs. A customer might say "Hey Alexa, turn off the kitchen lights," another customer may tap on the GUI, and another may utilize a remote. The input changes based on context but the object, action, and expected response will remain consistent to customers.

10 · From screens to objects

From screen based flows to object based systems

The object model reduced reliance on device specific flows by giving teams a shared way to define interaction behavior. Instead of designing every surface as a separate experience, teams could reuse object definitions and adapt them to the device context.

11 · Example in product

Example in product

A customer wants to turn off the kitchen lights while watching TV. In a standard design system with a screen based model, the flow depends on where the customer starts. In an object model, the system identifies the object, exposes the correct action, renders the state change, and confirms the result across the active surface.

12 · What this enabled

What this enabled

The model unlocked consistency, scalability, multimodal behavior, and AI readiness across the Alexa family.

13 · Outcome

Outcome

The object model created a clearer foundation for designing Alexa experiences across devices ahead of the Alexa+ launch. It helped to define the reusable patterns, reduce fragmentation, and support a scalable approach to multimodal interactions. As Alexa was moving toward a more dynamic and AI driven experience, this model provided a structure for thinking about how core capabilities could be discovered, acted on, and reused across the family of devices.

Screens describe where interaction happens. Objects define what interaction means.

Reflection

Object models are slow, careful work. The wins are quiet, the absence of friction, the meeting that ended in ten minutes instead of an hour. Worth every iteration.

Introduction

Fire TV is one of the largest TV platforms globally, used by millions of customers each day.

Context

Following years of incremental updates since launch in 2014, the platform had accumulated significant design and tech debt as we'd been quickly expanding to support new capabilities, monetization models, partnerships, and the ever evolving Alexa. When this project began, we essentially had a static launcher with a giant billboard of an advertisement above it. The customer experience was outdated and needed to grow into a unified content discovery experience to help customers find what to watch without knowing where it lives. This work is a direct result of the Object Model project I completed on Alexa.

The problem

Customers were navigating around an app first interface to find content that lived in many places. The architecture asked customers to know where things lived; we needed it to help them get to what they wanted to watch.

Approach

• Conducted rapid iterative testing (RITE) across Fire TV endpoints with eye tracking and biometric tools.
• Iterated live on emerging insights from the room next to the lab, refining navigation, labeling, and visibility each session.
• Validated concepts longitudinally through video diaries and incremental rollouts in select markets including Japan.
• Applied the configuration based model from the Alexa Object Model workstream to Fire TV.

Outcome

• A significant evolution of Fire TV emphasizing faster performance, simplified navigation, and intuitive content discovery.
• Reduced fragmentation and surfaced relevant content more effectively across services and capabilities.
• Aligned Fire TV with the broader Alexa and Echo platform to support a more seamless, scalable experience.

01 · Context

Where Fire TV stood

Fire TV is one of the largest TV platforms globally, used by millions of customers each day. After years of incremental updates since 2014, the platform had accumulated significant design and tech debt as we'd been quickly expanding to support new capabilities, monetization models, partnerships, and the evolving Alexa.

When this project began, we essentially had a static launcher with a giant billboard of an advertisement above it. The customer experience was outdated and needed to grow into a unified content discovery experience. This work is a direct result of the Object Model project I completed on Alexa.

02 · Research

What the research surfaced

To better understand content discovery and navigation behavior, we conducted rapid iterative testing (RITE) across Fire TV endpoints, exploring new information architecture concepts through live prototyping and continuous refinement with participants on the other side of the wall. Eye tracking and biometric tools were introduced into the test to better understand attention, engagement, and navigation behavior in real time with customers. The primary focuses were navigation structure, labeling, and visibility, which impacted users' ability to locate content and understand what the device offers.

The research revealed that many usability issues were not just caused by interactions, but by mismatches between system structure and user mental models. Customers consistently interpreted the menu as a space for system settings rather than a gateway to additional content, which led to low discoverability of capabilities such as games and audio. Adjustments such as increasing menu visibility and refining category naming improved both task success and awareness of available features.

Research continued to reveal what we already knew about our baseline: customers struggled to understand available capabilities and relied heavily on knowing where content lived. By iterating in real-time directly on emerging insights during the tests in the room next to the lab, we improved clarity, navigation speed, and overall awareness of what the experience offered. This resulted in consistently higher satisfaction and task completion success rates compared to the previous core experience as the weeks passed.

In addition to lab based testing, we incorporated longitudinal methods such as video diaries and validated our concepts through incremental rollouts across select markets, such as Japan. These findings reinforced the importance of aligning information architecture with user expectations, and directly informed additional work on structuring device experiences as systems of clearly defined objects, capabilities, and entry points.

03 · The goal

The goal

Shift Fire TV from an app first interface to a content first discovery experience to reduce time to content, unify content and monetization recommendations across services, simplify navigation, and support Alexa's expanding and evolving capabilities.

04 · The opportunity

The opportunity

Streaming ecosystems are inherently complex, with customers navigating their favorite streaming services like Prime Video, Netflix, and Hulu. Customers also are listening to music, changing their screensaver, and checking their security cameras. Customer behaviors from the RITE study showed that users often default to repeatedly opening the same app, even when better recommendations exist elsewhere. Customers often said things like "and when I click on it, 'aha' it's right there," despite the extra effort and time to ingress into the app.

There was an opportunity to reduce decision fatigue by shifting from app based navigation to a unified discovery experience to help customers quickly get to the content that they want to watch without needing to choose where to look first.

This was the time to also move towards a more flexible, configuration based model that came out of the Alexa Object Model workstream. This approach enabled experiences to surface core capabilities more dynamically, creating an opportunity to better unify Fire TV with the broader ecosystem and family of devices. This all ladders up to exposing relevant content and capabilities earlier in the experience.

05 · The solution

The solution

A content first discovery model was introduced that shifted the Fire TV away from app based navigation towards a unified, system driven experience that harmonized with the Echo devices and Fire OS tablets.

By restructuring the information architecture and surfacing content independent of its source, the experience enables faster discovery with clearer navigation and a more scalable integration for emerging capabilities.

06 · Exploration

What I did

We explored a persistent config switcher menu against a non-persistent treatment to find the right balance between content immersion and capability discovery. RITE testing showed that surfacing the menu persistently improved awareness of non-video capabilities like Games, Music, and smart home, participants formed a clearer mental model of the main screen for content and the menu for everything else ("aha, it's right there").

Even so, we ultimately went without the persistent config switcher. The persistent treatment carried tradeoffs for immersion, cross device consistency with Multimodal and Mobile endpoints, and the existing Menu icon's strong "settings" mental model. The research informed how we surfaced capabilities in the shipped, content first direction without locking system controls onto every screen.

06b · Exploration

What earned a place in the system

Beyond persistent vs. non-persistent, we explored what the config switcher itself should look like and contain. We tested icons versus no icons to see how customers parsed the row at a glance, and explored placement of Profiles, Settings, and the App Store, which capabilities lived inside the switcher versus elsewhere in the IA.

The through line was a definition exercise: what earned the right to be a config? We landed on a tight set of system level capabilities that customers needed to switch between quickly, while pushing one time or deeper tasks into their own destinations.

07 · Key decision

The key decision

Alongside the navigation work, we restructured how Settings and Profiles live in the information architecture, making them easier to find, easier to understand, and clearly distinct from content discovery. Profiles became a dedicated "Who's watching Fire TV?" entry point with recognizable avatars, and Settings was reorganized into a clear, scannable grid of categories so customers could orient quickly without hunting through nested menus.

06 · Experience architecture

The experience architecture

The architecture was designed as five layers that work together to scale across capabilities and partners, and to align Fire TV with the broader Alexa platform.

The goal

Shift Fire TV from an app first interface to a content first discovery experience to reduce time to content, unify content and monetization recommendations across services, simplify navigation, and support Alexa's expanding capabilities.

The opportunity

Streaming ecosystems are inherently complex, customers juggle Prime Video, Netflix, and Hulu while listening to music, changing screensavers, and checking security cameras. RITE research showed users default to repeatedly opening the same app even when better recommendations exist elsewhere. Customers often said things like "and when I click on it, 'aha' it's right there," despite the extra effort to ingress into the app. There was an opportunity to reduce decision fatigue by shifting from app based navigation to a unified discovery experience.

The solution

A content first discovery model that shifted Fire TV away from app based navigation toward a unified, system driven experience that harmonized with Echo devices and Fire OS tablets. By restructuring information architecture and surfacing content independent of its source, the experience enables faster discovery with clearer navigation and a more scalable integration for emerging capabilities.

Experience architecture

• Content First Discovery, surfaces recommendations independent of their source so customers don't have to know where content lives.
• Unified Content Layer, a single discovery layer with aggregated content, ranked based on what we know about the customer.
• Simplified Navigation Layer, reduces depth and cognitive load to enable faster movement between content, features, and destinations.
• Dynamic Configuration Layer, flexible configuration based model that scales and aligns Fire TV with Alexa and Echo platforms.
• Interaction & Focus Model, refined patterns and focus behaviors for faster navigation and clearer orientation, voice or remote.

Reflection

Architecture work is mostly negotiation. The diagram is the easy part, building the agreement that protects it is the work.

Introduction

Kroger is one of the largest retailers in the United States, serving millions of customers each day across thousands of stores and digital touchpoints.

Context

Kroger launched a major brand transformation in partnership with DDB Worldwide, introducing a new logo, typography, color system, and brand characters ("Krojis") as part of the "Fresh for Everyone" campaign. This transformation was designed to unify Kroger's identity across physical and digital experiences and differentiate the brand in an increasingly competitive grocery market. Shortly after, the COVID 19 pandemic accelerated a massive shift toward online grocery, pickup, and delivery, creating new digital touchpoints at unprecedented speed. The new brand existed but there was no clear system for applying it consistently across digital products, and the existing design system was engineering led and very technical, without a lot of flexibility for brand value as the modality for purchasing groceries shifted from in-store to online.

The problem

The brand transformation created misalignment between brand guidelines and product experiences. Product teams struggled to translate the new identity into usable UI patterns, leading to inconsistent layout systems and spacing, fragmented component implementations, unclear usage of typography and color, and inconsistent application of brand assets (Kroji characters and food photography). An audit of Kroger's digital ecosystem confirmed that inconsistency existed across web, mobile, and the associate experiences.

Approach

• Audited the existing digital ecosystem across customer and associate experiences.
• Translated DDB's brand transformation into product ready tokens, components, and usage rules.
• Ran desirability testing on the spectrum from transactional (Walmart like) to experiential (Target like).
• Stood up a federated governance model with reviews, office hours, and on-call support.
• Launched an internal podcast to teach the system, explain decisions, and interview partners.

Outcome

• Shipped the Kroger Digital Style Guide as a shared foundation for product teams across web and mobile.
• Enabled consistent application of the new brand as Kroger scaled pickup and delivery during the pandemic.
• Reduced inconsistency and improved iteration speed instead of slowing teams down.
• Established WCAG AA (4.5:1 contrast) as the accessibility baseline across the system.

Principles

• Brand values translate to product decisions, not decoration.
• Accessibility is a baseline, not a feature.
• Federated governance over a single gatekeeper.
• Teach the system out loud, reviews, office hours, podcast.
• Clarity and efficiency for the cart; brand expression for the journey.

02 · The opportunity

The opportunity

An audit of Kroger's digital ecosystem confirmed inconsistency across web, mobile, and the associate experiences, fragmented components, unclear typography and color usage, and inconsistent application of Krojis and food photography.

The convergence of a brand transformation and a rapid shift to digital grocery created a clear opportunity to define a system that translates brand into product while scaling across experiences.

03 · The goal

The goal

Create a centralized digital style guide and governance model that enables teams to consistently apply Kroger's brand across product experiences while supporting rapid growth in pickup and delivery services.

04 · System foundations

The system foundations

Four foundations the rest of the Kroger system is built on.

05 · The solution

The solution

The Kroger Digital Style Guide turned brand guidelines into a usable, end-to-end design system.

06 · The challenge

The challenge

This work took place during the pandemic, when digital grocery demand surged, teams launched features constantly, and consistency and speed were always in tension. The challenge was to create structure without slowing teams down.

07 · Enablement

What I did

A system only ships if people use it. Enablement was as much of the work as the system itself.

08 · Desirability testing

Desirability testing

Hypothesis: a more 'experiential,' brand forward direction (closer to Target) would feel more desirable than a utilitarian, efficiency driven one (closer to Walmart), and would give Kroger a clearer point of differentiation in a category where competitors look increasingly alike.

To test it, we ran a desirability study across Kroger, Target, and Walmart screens. Participants rated visual appeal on a 7 point scale, described what was and wasn't appealing in their own words, and selected from a fixed list of desirability attributes (Fresh, Stale, Helpful, Frustrating, Fun, Boring, Innovative, Dated, and so on) so we could compare reactions across designs on the same vocabulary.

Finding #1: visual appeal scores were nearly identical across Kroger (5.8), Walmart (5.68), and Target (5.66). The 'experiential vs. transactional' frame mattered far less than we assumed, customers did not strongly prefer one end of the spectrum over the other.

Finding #1a: the more 'transactional' home screen actually elevated Fresh and Fun, driven by vibrant food imagery and clear merchandising moments like Watermelon Season. Brand personality was coming from the photography and content, not from how lifestyle oriented the layout felt.

This reframed the system. Instead of choosing a side of the spectrum, the style guide prioritized clarity and efficiency for core shopping tasks and used food photography, color, and Krojis as the primary carriers of brand expression.

08b · A/B testing variants

A/B testing variants

Once the system was in market, the same kinds of questions came up at the component level. For a holiday hero on the homepage, we tested four variants of the same module: identical hero imagery and 'Set to Celebrate' messaging, but four different ways to surface the related product cards.

Variant 1 keeps the hero clean and lets the product rail live below it. Variant 2 floats the cards over the bottom of the hero. Variant 3 pulls them into a single 'Get the Ingredients' bundle. Variant 4 stacks taller cards directly over the imagery.

We ran these as A/B tests with small segments of customers and as moderated desirability sessions, asking the same kinds of questions from the earlier study, what feels Fresh, Helpful, Convenient, Cluttered, Hard to Use. The variants gave us a way to isolate one decision (how merchandising attaches to the hero) without changing brand expression, and to let real behavior and customer language pick the direction instead of internal preference.

09 · Outcomes

The outcome

The Digital Style Guide became a shared foundation for Kroger's product teams, enabling more consistent and scalable experiences across web and mobile. As Kroger expanded pickup and delivery, the system helped teams apply the new brand consistently across rapidly evolving customer experiences. Rather than slowing teams down, it improved alignment, reduced inconsistency, and supported faster iteration. The project demonstrates how a design system can fill the gap between brand and product while enabling teams to scale during rapid change.

Reflection

Systems live and die by adoption. Pretty documentation does not matter if the components do not show up in production. The podcast, the office hours, the ~100 design reviews, that's where the system actually shipped.

Introduction

Fire TV serves tens of millions of customers, and the companion app is used by millions more. With this many users, even small points of friction in set up can impact a massive number of customers.

Context

First impressions matter, on a date, in a job interview, and even more when someone turns on a device for the first time. 73% of Fire TV customers say their purchase decisions are influenced by their initial product experiences, with ease of set up being a major factor. Historically, Fire TV setup occurred entirely on the primary device using a remote control. While this supported basic activation tasks, it created significant friction for activities requiring text input or multiple decisions, Wi Fi configuration, account authentication, permissions, and personalization. Internal research showed customer behavior has shifted toward multi device interaction patterns: TV viewing rarely happens in isolation, and over 70% of customer time was spent on input heavy steps like account auth or address input. There was a mismatch between the complexity of the task and the device used to complete it.

The problem

Setup behaved like a fixed, linear flow on the worst possible input device for the most input heavy tasks. Customers described it as slow, confusing, and overly procedural, many were unsure whether setup was complete or what information had been collected. Thousands of customer service calls came in about set up alone. The system needed to communicate progress, build trust, and use the right device for the right task.

01 · Context

Where Fire TV stood

This is the moment customers decide whether the product feels intuitive, trustworthy, and worth the investment after unboxing.

02 · Research

What the research surfaced

Product requirements initially scoped this as porting the existing flow into a new design system, no changes to the flow itself. Coming off the experience architecture work, I knew that wasn't enough. I synthesized internal usability studies, competitor audits, and behavioral research on device onboarding.

The most time consuming and cognitively demanding parts of setup were tasks requiring text entry or repeated authentication across devices. Customers described setup as slow, confusing, and overly procedural, many were unsure whether setup was complete or what information had been collected. Thousands of customer service calls came in about setup alone. Beyond simplifying interactions, the system needed to communicate progress and build trust.

03 · Discovery

The discovery

I audited onboarding flows across Google TV, Roku, Apple TV, and Amazon's own multimodal devices. Google provided the most structured second screen onboarding with strong progress feedback and continuity. Apple excelled at ecosystem level automation through account and device synchronization. Roku offered the broadest accessibility through browser based setup but felt transactional rather than experiential.

The opportunity was to combine the strengths of all three into a more integrated dual screen experience. I also led a workshop with product and engineering partners in Berlin to align on business requirements, feasibility, and tech advancements.

04 · Reframing the problem

Reframing the problem

Early exploration showed the challenge wasn't simply redesigning setup screens. The existing experience treated onboarding as a linear sequence on a single device, even though tasks varied widely in complexity. Setup feels long not because every step is slow, but because a few high effort tasks (account registration, password entry, address input) dominate the process when done with a remote.

In reality, setup behaves as a conditional system: some steps depend on others, some can be skipped, others can be deferred. Treating it as a fixed step by step flow creates unnecessary friction. Traditional progress bars become misleading because the number of steps changes based on user choices.

Customers already engage in second screen behavior, phones in hand while interacting with the TV. Introducing a companion device during setup wouldn't require new behavior, just build on something already natural.

05 · Aligning scope with customer needs

The goal

Business requirements led us to initially scope this as rebuilding the entire OOBE flow into mobile to offer the most complete setup on market. Had we gone through with that, we would have ported over the existing linear flow, and all of the known customer pain points, into a different viewport. The business would have been happy; customers would have been just as frustrated.

I readjusted to prioritize speed to content and leverage new technology. This would have essentially been a UI redesign had I not pushed for it as a larger level of effort.

06 · Adaptive flow

The adaptive flow

An adaptive flow meets customers where they are by adjusting to their needs, context, and level of familiarity, rather than forcing everyone through the same rigid sequence.

In a linear flow, all users, regardless of experience, intent, or setup conditions, complete the same tasks in the same order. An adaptive flow responds to user input, device state, and prior information, letting experienced users accelerate while giving others the guidance they need. This reduces cognitive load, avoids redundant work, and feels more respectful of the customer's time, leading to faster, more confident completion.

07 · Object model

The object model

The system maps onboarding tasks across five primary objects, each representing a group of related configuration actions that contribute to the customer's final device state. The diagram shows how parts of the system depend on each other and why certain steps must happen before others. Device must be powered on before access. Access enables preferences. All of these feed into monetization, but only after certain conditions are met.

08 · Orchestrated interaction model

The orchestrated interaction model

The new system treats the television as a reassuring surface that communicates high level progress, while the companion device handles detailed interactions and data entry.

Sensitive information, account credentials, forms, consent, stays on the personal device, reducing privacy concerns associated with displaying it on a shared screen (an important use case because a preteen can eyeball your credit card details for their future Minecraft purchases). The TV stays visually calm: short status updates like "Connecting," "Signing in," or "Finalizing," while the companion device manages the underlying tasks. Both devices stay synchronized without duplicate information or conflicting instructions.

10 · Outcome

The outcome

The resulting system establishes a mobile first onboarding model that distributes setup tasks across devices while keeping the experience cohesive. Moving complex configuration to the companion device reduces the friction of remote based input and shortens the path to content.

Beyond improving setup itself, the architecture creates a foundation for faster iteration. Decoupling key interactions from the television interface lets teams test new experiences and ship improvements without waiting for full operating system updates.

Looking back, a few things would have made the work sharper: spotting the shift in effort sooner, partnering with visual design earlier on density and alignment, and bringing UX writers in from day one rather than leaning on marketing copy. Even so, the project reframed device setup from a static onboarding flow into a coordinated interaction system, and that shift is what I'm proudest of.

The goal

Make Fire TV the fastest set up experience by prioritizing speed to content instead of a linear CX that completes every configuration step upfront.

Reflection

Onboarding is the rare design problem where the goal is to be invisible. If customers remember it, you have probably done too much.

Introduction

Amazon runs one of the largest logistics networks in the world, where operators make fast, high-stakes decisions in real time. As AI rolled out across these workflows, the system needed clear, consistent guidance to keep that pace trustworthy at scale.

Context

AI capabilities were expanding across transportation products faster than the patterns supporting them. Without a shared framework, every team was inventing its own answer to the same questions, how AI should behave, where it should appear, and when it should stay out of the way, leaving experiences fragmented and hard to trust.

The problem

There was no unified system for how or when AI should assist. Behaviors drifted across products, interaction patterns were inconsistent, and there were no shared rules for when AI should guide versus stay passive. Underneath all of it sat a deeper gap: no framework for integrating AI into complex, table-based operational workflows in a way that supported clarity, trust, and confident decision-making.

Approach

• Defined the roles AI could play in a workflow, prompting, recommending, summarizing, or assisting step by step.
• Built a reusable set of interaction patterns for how AI appears, speaks, and behaves inside operator tools.
• Placed guidance directly in the flow of work, so assistance arrives at the right moment without interrupting the task.
• Wrote trust and clarity principles so every AI output is understandable, actionable, and appropriately framed.
• Standardized message anatomy around three parts: a clear label, a concise explanation, and a direct action.

Principles

• AI supports the workflow, it doesn't interrupt it.
• Guidance is contextual and intentional.
• Recommendations are clear and explainable.
• Patterns are reusable across transportation surfaces.
• Consistency is how trust compounds over time.

01 · Context

Where the platform stood

Amazon's transportation network, fulfillment centers, sortation, delivery stations, runs on time-sensitive workflows where operators make fast, accurate decisions in real time. As AI capabilities expanded across these products, every team was shipping its own patterns, with no shared framework for how AI should behave or appear.

02 · Challenges

The challenge

The system had to stay flexible enough for teams to keep innovating, but consistent enough that customers experienced one coherent AI across every brand, interaction, and screen size. Moving at the speed AI was emerging meant making decisions before every edge case was known, and designing patterns that could absorb change without fragmenting.

03 · Design decisions

Message anatomy, label, explanation, action

Every AI message follows the same three-part structure so operators know what they're looking at, why it matters, and what to do. The shape stays consistent across informational, warning, and critical types; only the visual weight changes to match urgency.

04 · Design decisions

Border, contrast, and corner radius

Borders and contrast were tuned so AI surfaces feel distinct without shouting, and corner radius explorations landed on a value that read as calm and modern across both brands and screen densities.

05 · Design decisions

Theme explorations and avatar variations

Theme studies covered light, dark, and high-contrast modes so AI guidance stays legible across operator environments, from warehouse floors to dispatch monitors. Avatar variations explored how the AI presents itself, friendly enough to feel approachable, neutral enough to stay out of the way.

06 · Design decisions

Chatbot window and the relay assistant

The chatbot window pattern docks AI inside operator tools without taking over the workspace, and the relay assistant pattern hands work between AI and human at the right moment, with full context preserved.

07 · The opportunity

The opportunity

Without a unified system, behaviors drifted across products: fragmented interaction patterns, no clear rules for when AI should step in or stay quiet, and a real risk of eroding trust in a high-stakes operational environment. The deeper gap was a missing framework for integrating AI into complex, table-based workflows.

08 · Demo

The demo

A live look at the chatbot agent docked inside an operations dashboard, showing the message anatomy, suggested actions, and the calm, contextual cadence the framework defines.

09 · Design guidance

Design guidance

Every pattern shipped with a guidance page in the system so teams could pick it up and apply it without re-asking the same questions, usage, types, do's and don'ts, accessibility notes, and code, all in one place.

The opportunity

AI was being introduced product by product, with no shared definition of when it should help, how it should speak, or what it should look like. The opportunity wasn't another feature, it was a framework, one that could integrate AI into complex, table-based workflows while preserving clarity, trust, and operator focus.

The outcome

A shared system of AI guidance principles, interaction patterns, and behavioral roles that teams across transportation could build on. Rather than treating AI as a standalone feature, the system weaves guidance into the flow of work, making it clear when AI should recommend, prompt, summarize, or simply stay quiet. Messages follow a consistent anatomy of label, explanation, and action; tone stays clear and brief over conversational; and message types, informational, warning, and critical, each have distinct visual treatments tuned to urgency. Layout, accessibility, content do's and don'ts, interaction states, and escalation patterns are all part of the same kit, so teams can ship consistently without re-solving the basics.

Reflection

The hardest part of AI design isn't the model. It's deciding what the AI should refuse to do.

Introduction

A foundational focus and interaction system for widget based experiences across Amazon Alexa and Amazon Fire TV.

Context

I contributed to the foundational focus and interaction system for widget based experiences across Amazon Alexa and Amazon Fire TV. This work supported the introduction of widgets as a new, modular surface for glanceable and lightweight interaction within a 10 foot, non touch environment.

Unlike traditional apps, widgets are embedded, partial experiences. Customers do not "enter" a full interface. They navigate within layered, nested UI structures using a remote, voice, or hybrid input. There is no cursor or touch affordance, so focus becomes the primary system for communicating location, hierarchy, and action.

This project defined how focus works at scale across widgets from entry, to internal navigation, to exit, creating a consistent interaction model aligned with Alexa Presentation Language (APL) and the Alexa Design System.

The problem

Widgets are new and introduced a new level of structural complexity without an interaction model to support it. There was no system defining how focus should behave across these layers, and multiple teams owned different parts of the widget. Focus rings on the container were owned by a different team than the content inside.

Approach

• Customers needed to enter a widget from a broader surface.
• Navigate between internal regions like header, lists, and tiles.
• Interact with specific items.
• Exit back to the parent layout.

Outcome

• A hierarchical focus model: widget to subcomponent to item.
• Defined subfocus progression within widgets.
• Scroll linked focus behavior.
• High visibility focus states.
• Clear entry and exit rules.

Principles

• Focus is hierarchical, not flat.
• Anchor entry to structure, not content.
• Focus stays stable as content scrolls.
• One model across remote, voice, and keyboard.
• Visibility wins at a distance.

01 · Overview

A new surface for Alexa and Fire TV

I contributed to the foundational focus and interaction system for widget based experiences across Amazon Alexa and Amazon Fire TV. This work supported the introduction of widgets as a new, modular surface for glanceable and lightweight interaction within a 10 foot, non touch environment.

Unlike traditional apps, widgets are embedded, partial experiences. Customers do not "enter" a full interface. They navigate within layered, nested UI structures using a remote, voice, or hybrid input. There is no cursor or touch affordance, so focus becomes the primary system for communicating location, hierarchy, and action.

This project defined how focus works at scale across widgets from entry, to internal navigation, to exit, creating a consistent interaction model aligned with Alexa Presentation Language (APL) and the Alexa Design System.

02 · The problem

Structural complexity without an interaction model

Widgets are new and introduced a new level of structural complexity without an interaction model to support it.

Multiple teams also owned the widget. Focus rings on the container were owned by a different team than the content inside the widget.

Internal research showed that users perceive remote interaction as effortful and click heavy, often preferring voice when possible. We also discovered this during the setup portfolio piece. That means any inefficiency or ambiguity in navigation, especially around focus, has a big impact on usability.

Research also showed that customers struggled to identify what was selected at a distance. Focus was often the only signal of interaction state. Without a defined system, we would launch something unpredictable, visually unclear, and inefficient.

03 · Research

What remote navigation costs the user

I leveraged usability testing on remote navigation patterns across Fire TV experiences. The research highlighted several key constraints.

04 · Gaps identified

What was missing

The most significant gap was the absence of a focus model. Widgets contain nested interactive elements, but there was no system defining how focus transitions between layers.

There was also no concept of subfocus, or how focus behaves within a component. For example, text list widgets required a clear progression from header to list items, but this interaction pattern had not been formalized.

Finally, there was no defined system for scroll plus focus interaction, which is critical for list based widgets. Without it, navigation could feel disjointed or visually unstable.

05 · Opportunity

A foundation, not just consistency

This created an opportunity to define a foundational interaction model for widgets from the ground up.

The goal was not just consistency. It was reducing interaction cost in a constrained input system.

There was also an opportunity to align interaction behavior from Fire TV with the APL design system team I was on, enabling reusable patterns that scale across widget types without redefining logic for each experience.

06 · Design approach

A hierarchical focus model

First, I introduced a hierarchical model with three levels.

Next, I defined subfocus progression, which governs how users move within a widget. For example, in text list widgets, focus enters at the header, moves sequentially through list items, and maintains directional consistency.

07 · Key design decisions

Decisions for focus states

A small set of decisions kept the model coherent across teams and surfaces.

08 · Scroll plus focus

Scroll linked focus behavior

Scroll and focus had to behave like one motion. As the user presses down through a list, the focused item stays put while content advances behind it. When they reverse direction, focus rides the list back to the top, never losing its anchor.

09 · Solution

A foundational focus and interaction system

The result was a foundational focus and interaction system that defines how users enter, navigate within, and exit widgets across Alexa and Fire TV.

Within APL, this translates into reusable patterns for defining focusable elements, interaction states, and navigation behavior, enabling consistent implementation across widget types.

10 · Visual specs

Making focus visible at ten feet

Focus needed to read clearly from across a living room. The container state used a high opacity gradient border so it would never read as a hover. The item state used a soft inner fill so the eye could land on a single line in a dense list. Together they gave the user a stable, legible cue regardless of distance or content density.

11 · Outcomes

A scalable interaction foundation

This work established a scalable interaction foundation for widgets as a new surface area.

For customers, it improved navigation clarity and reduced effort, particularly in complex or content dense layouts. By making focus highly visible and behavior predictable, users could navigate more confidently, even at a distance.

For teams, it reduced ambiguity and enabled faster development by providing a shared interaction model aligned with APL. This prevented fragmentation as widgets scaled across different experiences.

Reflection

Focus is the cursor of a 10 foot, non touch world. Get it right and the whole surface feels calm.

Introduction

The Smile Advisor, an AI oral care expert that curates a routine and gets users one step closer to a healthier smile.

Context

Around 2018, Crest was expanding its digital presence as more consumers researched and purchased personal care products online rather than in store, alongside a growing direct to consumer marketing motion. At the same time, the at home whitening category had become increasingly crowded, with multiple product variations that were difficult for customers to differentiate. Without an easy way to assess their own needs, many users relied on guesswork when choosing a product. The Smile Advisor app was created to provide personalized recommendations, helping customers understand their current tooth shade and confidently select the right whitening solution.

The problem

Oral care can be an overwhelming world of information, especially after a trip to the dentist. With all the different products and health considerations, it gets tricky, and it can also be a sensitive area with insecurities. Many of us have a bathroom shelf dedicated to the trial and error of new products, and customers were skeptical about risking the health of their teeth, money, and time to experiment.

Approach

• Partnered with the P&G Oral Care team to map customer questions a hygienist would actually ask.
• Wireframed a short selfie + Q&A flow with Mindtree engineering.
• Iterated through many UI variations and multiple rounds of design review on site with the client.
• Ran A/B testing on site with the client across numerous days.
• Worked alongside engineering to develop the interactions and interface to match the prototypes.

Outcome

• Transformed product selection from a static, guesswork driven purchase into a personalized, interactive journey.
• Increased customer confidence in product choice and reduced uncertainty during decision making.
• Created a more engaging path to purchase by connecting product education directly to the user's needs.
• Demonstrated how simple, user driven inputs can make complex product decisions feel clear, intuitive, and tailored.

Principles

• Helpful, not pushy.
• Show the why, not just the what.
• Make the sensitive feel safe.
• Visual input lowers the barrier.
• Trust earns conversion.

01 · The goal

The goal

Ease an overwhelming, sensitive decision and increase conversion by turning product discovery into a personalized, guided experience. With a selfie and a few questions, give users an AI expert analysis and a personalized routine that builds confidence on the way to purchase.

02 · Context

The category problem

Around 2018, Crest was expanding its digital presence as more consumers researched and bought personal care online. The at home whitening category had become crowded, with multiple SKUs that were hard to differentiate, and most users were guessing.

03 · Demo

The demo

A walkthrough of the app, selfie capture, shade analysis, the short Q&A, and the personalized regimen at the end.

04 · The flows

The flows

Built with Mindtree engineering and the P&G Oral Care team, wireframes led to many UI variations and multiple rounds of design review, both on site and with the client.

05 · A/B testing

Testing and validating

Iterated on site with the client over numerous days, testing UI variations side by side to refine tone, layout, and the moments where customers needed reassurance most.

06 · Outcome

The outcome

Smile Advisor turned product selection from guesswork into a personalized, interactive journey, building confidence in product choice and connecting education directly to the user's needs.

This was one of my first projects as a designer, and I'm still grateful to the oral care brand managers at P&G, the engineering team at Mindtree, and Publicis Groupe. Crest remains my favorite brand I've worked on.

Reflection

Trust compounds. Earn it in the first interaction or you do not get a second one.

Introduction

Studio by Tide, a new sub brand that moved Tide from purely functional messaging into a more lifestyle driven relationship with the customer.

Context

Tide is one of the most recognized laundry brands in the world, operating in a category used by nearly every home in the country. Studio by Tide marked a shift in how the brand connected with customers, moving from purely functional messaging to a more lifestyle driven approach.

The problem

The campaign had to introduce a new product line while building awareness, driving engagement, and creating a clear path to purchase across email and social, all while keeping the everyday Tide voice intact at the edges.

Approach

• Designed a series of CRM emails grounded in user research for the P&G customer relationship management program.
• Worked across clients, copywriters, data strategists, and developers to keep the system aligned end to end.
• Built the key visuals that other designers extended across email and social placements.
• Targeted users into product specific buckets based on the feedback they gave inside the emails.
• A/B tested live with the client over numerous days to refine tone, layout, and interaction.

Outcome

• Above average open and click through rates across the email program.
• Stronger reach and brand visibility from social placements on Facebook and Instagram.
• Increased first time purchase intent and a more direct path from awareness to conversion.
• The campaign became a basis for future Tide testing programs.

Principles

• Lifestyle, not lecture.
• Every email is also a survey.
• Personalization earns the next click.
• Ship the system, not the screenshot.

01 · Context

The new brand

Tide is one of the most recognized laundry brands in the world, operating inside a category used by nearly every home in the country. Studio by Tide was a new sub brand that marked a shift in how Tide connected with customers, moving from purely functional messaging to a more lifestyle driven approach.

02 · The challenge

The challenge

The campaign needed to introduce the new product line while building awareness, driving engagement, and creating a clear path to purchase across email and social channels. The voice had to feel new without feeling unfamiliar.

03 · The role

The role

My role was to create a series of emails grounded in user research for the P&G CRM (customer relationship management) program. I worked between the clients, copywriters, data strategists, and developers, and built the key visuals that other designers extended across the campaign.

04 · The emails

The emails

This was the initial launch of a new product series and the start of a new consumer journey for Tide customers. The emails were interactive and targeted to new consumers of Studio by Tide, framing the product line as part of a lifestyle rather than a spec sheet.

05 · The strategy

The strategy

Depending on the feedback we received through the emails, we targeted the user for a specific bucket of products tailored to their needs. The email was the survey, the segmentation engine, and the storefront entry point at the same time.

06 · A/B testing

Research and Validation

Iterated on site with the client over numerous days while working with engineering to develop the interactions and the interface to match the prototypes. Tone, layout, and the timing of CTAs were all tuned against live response.

07 · Outcome

The outcome

The Studio by Tide campaign successfully translated a lifestyle driven brand into a cohesive digital experience across email and social.

The email program drove strong engagement with above average open and click through rates, while social placements on Facebook and Instagram increased reach and reinforced brand visibility across key audiences. The brand saw increased first time purchase intent and a more direct path from awareness to conversion.

Overall, the campaign demonstrated how aligning messaging across channels can increase engagement, strengthen brand perception, and drive measurable action. We used this as a basis for future Tide testing programs.

Reflection

Lifestyle messaging only works when the system underneath it is boring and reliable. Build the rails, then write the story.

Introduction

Multi year creative work for two of the leading oral care brands in the world, spanning CRM emails, Times Square billboards, Hulu spots, Instagram Stories, gifs, Snapchat games, and the Whitestrips Glam campaign.

Context

Oral B and Crest are two of the most recognized oral care brands globally, trusted by millions of customers and widely recommended by dental professionals. I worked on P&G Oral Care for multiple years, creating digital campaigns for many media forms including Times Square billboards, Hulu advertisements, Instagram Stories, gifs, Snapchat games, and more.

Approach

• Ran A/B tests across Oral B email campaigns on the same P&G CRM platform used for Tide.
• Tested quiz style emails, simplified messaging, and scannable list based content against open rate, click through, and conversion.
• Partnered with Facebook/Instagram NYC, P&G, and Disney on story carousel and newsfeed ads for the Oral B Toy Story and Star Wars launches.
• Designed Hulu, Facebook, and Instagram spots for junior and teen brushes, testing copy, speed, content order, and music.
• Led the Not Just a Pretty Brush social campaign with multiple cutdowns built from a single visual system.

Outcome

• Reducing cognitive load and presenting information in more digestible formats consistently improved engagement.
• Structured, guided email experiences helped customers make decisions more confidently.
• Findings fed ongoing optimization of layout, CTAs, and messaging strategy across the brand.
• Co branded launches with Disney scaled cleanly across formats from story to newsfeed to broadcast.

01 · Context

The work

Oral B is one of the leading oral care brands globally, trusted by millions of customers and recommended by dental professionals. Over several years on P&G Oral Care, the work covered Times Square billboards, Hulu spots, Instagram Stories, gifs, Snapchat games, and more.

02 · The work

The work

A multi year body of creative work for Oral B and Crest, spanning short form social, Hulu spots, Facebook and Instagram story and newsfeed placements, Disney co branded launches with Toy Story and Star Wars, the Whitestrips Glam campaign, and CRM emails. Built from shared visual systems with many cutdowns, the work was tuned for silent autoplay, swipe behavior, and long form streaming.

03 · Outcome

The outcome

Over several years, the work moved Oral B and Crest from spec-driven product messaging into a more human, structured, and testable creative system that scaled cleanly from CRM inboxes to story carousels to broadcast.

Reflection

The product is rarely the hero. The customer's experience of the product is. Build a system flexible enough to prove that across every channel.