Back to Work
Case Study 02

Embedding AI into the workflow — not beside it.

Genie — EY's ambient AI layer, bringing GenIAus's intelligence to the exact moment auditors need it.

ClientErnst & Young
RoleProduct Designer
TeamVP Product · Motion · Dev
Year2023–2024
Genie — embedded AI widget inside the VIA platform
The Problem

GenIAus worked, but auditors wouldn't leave their page to use it — every query meant a separate tool and lost context.

My Approach

Rebuilt AI as an ambient, embedded layer — a contextual widget that comes to the user instead of the other way around.

−60%Context-switching between tools
~45sTime-to-first-query (was 4 min)
72%Selected sources before prompting
58%Used Page Awareness in sessions
+28%Increase in trust rating

GenIAus worked. Nobody wanted to leave their page to use it.

Auditors valued GenIAus — but wouldn't leave their page to use it.

A destination became an ambient presence.

Genie wasn't a feature addition to GenIAus — it was an architectural rethinking of how AI should relate to the user's workflow.

GenIAus was a destination — you go to it. Genie was an ambient presence — it comes to you. That distinction drove every design decision.

I worked directly with the VP of Product to define the product strategy across three pillars:

Technology. Establishing experience design operations and customer insights frameworks that could inform AI behavior across multiple tools — not just audit.

People. Designing for user trust, agency, and control in a domain where AI skepticism runs high and professional liability is real.

Business. Aligning Genie with EY's broader digital transformation — creating a scalable AI interaction layer that could eventually serve as the foundation for AI-assisted workflows across the entire VIA ecosystem.

Technology People Business Vision & Strategy Outcome 01 Experience Design Operations Outcome 02 Customer Insights & Metrics Outcome 03 Leadership & Governance Outcome 04 Customer Focus as Culture
Fig 01Product strategy framework — Technology, People, and Business as interconnected pillars around a shared Vision & Strategy core. Each pillar carried specific design operations and governance requirements that shaped Genie's architecture.

A chat panel on the side of the screen. Deceptively simple.

A chat panel sounds simple — three constraints made it anything but.

Dense information in limited space

The Risk Monitoring dashboard is already information-heavy — donut charts, heat maps, risk tables, entity breakdowns. An AI panel had to complement that density, not add to it.

Long response times

Genie queried multiple sources at once — document stores, help centers, internet, live page content — slower than a typical chat exchange.

Multiple contexts of use

A generic answer, a page-aware answer, and a task-embedded action are three different products, not one.

Surface. Context. Embedded.

I designed Genie around a layered integration model — three distinct contexts of use, each with increasing depth of contextual awareness.

More context, more depth — but always the user's choice how much AI is active.

Genie Widget Layer 01 · Surface Page Awareness Layer 02 · Contextual Deep Integration Layer 03 · Embedded What it does Global presence — accessible from every screen Generic Q&A on demand Conversation history within reach What it does Toggle on / off — user keeps control Reads & references the embedded page content What it does Genie sits inside the user flow itself Can assist & generate answers in-task Helps users complete real workflows end-to-end
Fig 02Three layers of integration — from global widget (surface) to page awareness (contextual) to deep integration (embedded). Each layer increases the AI's contextual depth while preserving user control over how much intelligence is active.
01
Genie Widget
  • The surface layer — a collapsible chat panel on any page in VIA
  • Generic Q&A, conversation history, prompt library access
  • Globally present, contextually unaware
02
Page Awareness
  • The contextual layer — when toggled on, Genie reads the current page
  • On Risk Monitoring: understands the 851 risks and their breakdowns
  • Queries become contextual — data-grounded answers, not generic ones
03
Deep Integration
  • The embedded layer — Genie becomes part of the user flow itself
  • Assists task completion, pre-fills content, guides within forms and workflows
  • The future state the architecture was designed to support

Seven decisions. One coherent system.

Each decision below was a deliberate response to a specific design problem — the constraint, the resolution, and the reasoning behind it.

7 problems, 7 targeted fixes — each preserving context, control, or trust.

01
The widget — always present, never intrusive.

The Problem

  • How do you make an AI assistant available on every page?
  • Without dominating the interface or distracting from the task

The Decision

  • Collapsible side panel anchored to the right edge
  • Collapsed: minimal orange tab with the Genie spark icon
  • Expanded: overlays the right portion at a fixed width, most content stays visible
  • Distinct visual identity (orange header, spark, chat bubbles) separates AI from verified system data

Why It Works

  • User stays on their page, context preserved
  • One click away, not one navigation away
  • Visual separation prevents AI output being confused with verified system data
Genie widget collapsed on the right edge of the Risk Monitoring dashboard
Fig 03The Genie widget on the Risk Monitoring dashboard — always accessible via the orange tab on the right edge, never intrusive. One click expands the full chat panel without leaving the page.
02
Resource selection before prompting — user agency over AI scope.

The Problem

  • Genie could query multiple sources — document stores, help centers, internet, live page content
  • Users in a regulated environment need to control where the AI pulls from
  • Querying everything by default was technically easy but experientially wrong

The Decision

  • "Select source" dropdown at the top of the chat panel
  • Checkboxes: Document store, Help Center, Internet, Page Awareness, Select all
  • Users choose sources before prompting, or change them mid-conversation

Why It Works

  • Gives users agency over what the AI sees
  • Narrows the source scope, which improves response time
Resource selection dropdown with checkboxes for Document store, Help Center, Internet, Data Query Observation, Select all, plus an Apply button
Fig 04Resource selection — users explicitly choose which data sources Genie queries. Not buried in a settings panel; a first-class interaction at the top of every conversation.
03
Page Awareness — AI that sees what you see.

The Problem

  • The most powerful use case: answering questions about what's on screen right now
  • But "page awareness" is a technical capability, needing a clear user-facing concept

The Decision

  • Page Awareness: a named, toggleable resource in the source dropdown
  • When enabled, a "Page Awareness Enabled" badge marks the conversation thread
  • Genie explicitly acknowledges the context shift in its own reply

Why It Works

  • Explicit beats silent for trust
  • Users always know when the AI is reading their page
  • The badge marks the before/after boundary in the thread
Page Awareness Enabled badge mid-conversation, with Genie acknowledging it has context about the current dashboard
Fig 05Page Awareness enabled — the AI explicitly confirms it has context about the current dashboard, and a visual badge marks the shift in the conversation. Users always know when the AI is reading their screen.
04
Source transparency — showing your work.

The Problem

  • In audit, the source of information matters as much as the information itself
  • A response without attribution is an opinion; with citations, it's a reference point

The Decision

  • Every response includes a collapsible Citations section
  • Shows source type and document names (e.g., Alpha Rep.pdf, +3 more)
  • Each citation is clickable — traces the AI's reasoning back to its source

Why It Works

  • Turns the AI from a black box into an accountable assistant
  • Auditors can verify claims against source documents — what their standards require
Source transparency — citations appearing inline with an AI response, listing Alpha Rep.pdf, Y_Report.pdf and others
Fig 06Source transparency — every response includes clickable citations showing exactly which documents from the Document Store informed the answer. Trust through traceability.
05
Dynamic resource switching — control without starting over.

The Problem

  • Users often realize mid-conversation they want a different data source
  • The initial model required starting a new chat — losing all context

The Decision

  • "Data Source" label shows which resources generated each answer
  • "Change Source" button switches sources mid-conversation without losing the thread
  • Next response uses the new sources; prior context stays preserved

Why It Works

  • Treats source selection as per-response, not per-session
  • Matches how auditors actually cross-reference within one analytical thread
Dynamic resource switching — a Change Source button under an AI response, with the source label visible above it
Fig 07Dynamic resource switching — users can change data sources mid-conversation without losing context. The Data Source label and Change Source button make source management a fluid part of the interaction, not a system setting.
06
Response control & feedback — the user stays in charge.

The Problem

  • Enterprise AI tools often present responses as final and authoritative
  • Audit professionals need to manipulate, verify, and refine outputs before use

The Decision

  • Action toolbar on every response: copy, compare, regenerate, share, export
  • Not hidden in menus — inline and visible
  • Prompt Library adds: Shorten, Spell Check, Find Similar

Why It Works

  • Positions the AI as a starting point, not an endpoint
  • Critical where AI output is an input to judgment, not a replacement for it
Response control toolbar — copy, compare, regenerate, share, export icons inline below an AI response
Fig 08Response control — copy, compare, regenerate, share, and export actions on every response. Plus a Prompt Menu for post-response refinement: Shorten, Spell Check, Find Similar.
07
Prompt library — task-specific intelligence.

The Problem

  • The embedded widget had even less room for prompt scaffolding than GenIAus
  • Users still needed guidance on what to ask

The Decision

  • Prompt Menu with task-specific actions, not generic suggestions
  • "Shorten" condenses, "Spell Check" reviews errors, "Find Similar" surfaces comparable entries
  • Operational prompts — they do things, not just ask things

Why It Works

  • In a compact widget, every pixel matters
  • One icon replaces four suggestion chips
  • "Find Similar" lets the AI learn from collective team behavior
Prompt menu in the widget showing Shorten, Spell Check, and Find Similar actions
Fig 09Prompt library in the widget — task-oriented actions (Shorten, Spell Check, Find Similar) rather than generic question suggestions. Optimized for the compact widget format.

Building the chat-focused AI library EY didn't have.

Designing Genie meant building the design system EY didn't have yet.

EY's design libraries had no components for conversational UI, AI states, citation patterns, or source selection. I built a dedicated AI/chat design system from scratch:

This design system wasn't just for Genie. It was designed to be the foundation for AI interactions across EY's entire product suite — portable enough to embed in any tool, consistent enough to build user familiarity across products.

Research that changed the product.

Shipping Genie's first iteration was only half the job — the harder question was whether Data Source Selection and Data Query Observation were actually understood. I ran remote moderated think-aloud sessions on Microsoft Teams with practicing auditors to find out.

Two users, ambiguous labels, one buried toggle — and the same confusion twice.

The research question: "How might we ensure that Genie's new features are intuitively understood and seamlessly integrated into the auditor's existing workflow?"

Screenshot of remote usability testing session — multiple researchers observing a participant interact with Genie inside VIA on a shared screen
Fig 10A remote usability session in progress — participants interacted with Genie's features on the live VIA platform while UX researchers observed behavior and captured insights in real time.
Participant 1
"What is Knowledge Store?"

"I think I understand what it is, but the label is unclear."

Participant 1
"Did I miss this earlier?"

"It's placed too low on the interface."

Participant 2
Internal or external?

"What are these sources? Are these internal databases or external ones?"

Both participants converged on the same root cause from different paths: ambiguous labels and a buried toggle were costing users time they were otherwise willing to spend exploring. Testing also surfaced three findings beyond the labeling issue:

Trust & Privacy

Users worried whether audit details were visible to others through the AI.

Internet Source

Users wanted its knowledge cut-off date and whether it covered current events.

First-Open Expectations

Users wanted a capability intro on first open, and to keep working on other pages while Genie processed.

Action-Thinks-Says journey map for two users across multiple session steps, with sticky notes in green, purple, and pink
Fig 11Action → Thinks → Says journey maps for two test participants. Each row tracks a user's complete session — every click, internal thought, and verbal statement — revealing where the design supported discovery and where it created confusion.
Empathy map with Think and Feel, Do, Say, Pain, and Gain quadrants populated with sticky notes
Fig 12Empathy map synthesizing the same testing sessions — Think & Feel, Do, Say, Pain, and Gain.
Thematic analysis board with six themes laid out as green column headers above blue sticky-note groupings
Fig 13Thematic analysis board coding the interview data into recurring patterns.

Three concrete changes. Test, redesign, ship.

The research findings drove three concrete design changes. The carousel below cycles through the iterations — from Design A and Design B explorations to the final design and the mid-conversation source switching pattern.

One label change removed every confusion observed in testing.

Renamed "Data Query Observation" to "Page Awareness." The original name was engineer-speak that confused every user tested; the new one describes exactly what the feature does.

Consolidated all data sources into one dropdown. Page Awareness joined Document Store, Help Center, and Internet in a single checkbox list — matching how users actually saw the decision.

Added source labels and descriptions, so users could choose without guessing whether a source was internal or external.

Embedded AI. Measurably adopted.

Five metrics, one story: embedding beat building a destination.

−60%Context-switching between tools — users no longer left VIA to access AI assistance
~45sTime-to-first-query, down from ~4 minutes with standalone GenIAus
72%Of users actively selected sources before prompting — strong adoption of the transparency model
58%Of all Genie sessions had Page Awareness enabled — validating the contextual layer
+28%Increase in user-reported trust rating, attributed to citations and source labels

Two questions. Two answers.

These two projects represent both sides of enterprise AI design — the standalone tool, and the embedded layer.

Build trust in a new tool, then carry that trust into a familiar one.

Case Study 01

GenIAus — build it.

Answered: Can we build an AI tool auditors trust? Yes — with the right onboarding, prompt scaffolding, and information architecture.

Case Study 02

Genie — embed it.

Answered: can we embed intelligence without breaking trust? Also yes — with a lighter touch and a layered integration model.

Together, they represent a complete AI design trajectory: from standalone to embedded, from feature-first to context-first, from building trust in a new tool to maintaining trust inside a familiar one.

What I'd do differently. What it taught me.

The system was the product — that's the one thing I'd protect first next time.

What I'd do differently

Measure response time by source configuration. The feature was designed partly to improve performance, but I lacked the instrumentation to prove narrower scopes actually reduced latency.

Prototype Deep Integration earlier. Only Widget and Page Awareness shipped in my tenure — starting the highest-value layer sooner would have de-risked it.

What it taught me

The best AI products don't feel like AI products — they feel like the tool you already use, but smarter. That's what "ambient" means.

Embedded AI is a design systems problem as much as an interaction design one. Without the component library, every future integration would start from scratch — the system was the product.

Next Case Study 03 — Peepal 9 million farmers. A $1 trillion market closing fast →

Have a project in mind?

I'm open to senior product design roles, advisory work, and selective collaborations. Whether you have a defined brief or a fuzzy problem space, let's talk it through.

damleaalvee@gmail.com
Let's
create.