Yaazh /
01 · Financial Engine 02 · AI Companion
CASE STUDY 02 · Enterprise Data & AI Systems · Product Definition Under Ambiguity

AI Data Companion

From “add AI” to designing an actual product.

The Starting Point
“Bring AI into the data integration experience.”
No Product Vision No established product model or clear functional boundaries.
No Future Journey No defined path of how an analyst would collaborate with AI.
Unformed Problem AI was prescribed as a requirement before the need was articulated.

AI was already mandated as part of the requirement, and I was asked to design the experience. But I didn't want to start by designing an AI feature — I wanted to understand what the user actually needed AI to help with.

01 — The Problem

We move data more than we realise.

Organizations constantly move data across systems for critical business initiatives:

01. Legacy Modernization
02. Cloud Migration
03. Regulatory Demands
04. Mergers & Acquisitions
05. Enterprise Data Platforms & Lakes

Moving data sounds straightforward: Take data from System A and put it into System B.

⚠️ But enterprise data rarely arrives neatly labelled. It is:
  • Inconsistently named across schemas
  • Missing vital business context
  • Scattered across disparate documents
  • Described differently across systems
  • Understood completely differently by different teams
The real challenge isn't simply moving the data. It's understanding the data well enough to connect it correctly.
02 — The Mental Model

The Moving-House Analogy

To anchor the design team and stakeholders around the real problem, I introduced a tangible physical metaphor:

The Data Integration Dilemma Interactive Conceptual Model
📦 SOURCE HOUSE 100s of Boxes
Companion Inspect ➔ Group ➔ Explain
🏡 TARGET HOUSE New Schema Rooms
Room: Corporate Loans Ledger 96% Match

Why it belongs here: Analyzed sample fields inside Box #104 — contains commercial credit facility identifiers, maturity dates, and counterparty entities matching Corporate Lending standards.

Analyst Role: Confirm mapping & update shared business glossary.
The Core Role of the AI Data Companion: Not an AI that simply spits out a brittle answer. An assistant that helps analysts build understanding and make trusted connections between systems.
03 — Who Is The User?

Meet Alia — The Data Analyst

👩🏽‍💻

Alia's Core Responsibility

Understand complex enterprise data, build trusted business knowledge, create accurate source-to-target mappings, and support multi-million dollar integration projects.

❌ What Success Is NOT

“AI magically generated a black-box mapping.”

✅ What Success ACTUALLY IS

“I understand this data well enough to trust the mapping I'm approving.”

Alia's 4 Criteria for Trusted Success:
1. Trusted Glossary

Business meaning is documented clearly, agreed upon, and reusable across teams.

2. Accurate Mappings

Source and target field relationships make complete technical and business sense.

3. Faster Delivery

Less manual grind spent discovering schemas and decoding obscure field names.

4. Explainable Recommendations

The analyst clearly understands the underlying reasoning behind every AI suggestion.

04 — The Existing Journey

Understanding Comes Before Mapping

Before designing any AI UI, I mapped what the analyst actually had to do step-by-step:

STEP 01 Receive Materials Files, schemas & docs arrive.
STEP 02 Explore Data Examine source tables.
STEP 03 Build Meaning Piece together definitions.
STEP 04 Create Mappings Link source to target.
STEP 05 Validate Check if mappings hold.
STEP 06 Approve Sign off outcome.
💡 The Critical Observation: A large part of the work happens before the actual mapping. Understanding comes first. If AI only generates mappings without building understanding, the analyst cannot trust or approve the output.
05 — Working Backwards

The Product Wasn't Defined Yet

There was an AI requirement, but there wasn't yet a clear product model around it. Instead of jumping directly into screens, I worked backwards through collaborative discovery:

Cross-PM Alignment

Interviewed multiple PMs across Data Ingestion, Catalog, and Integration to identify where AI could genuinely relieve cognitive load.

Storyboards & Interaction Models

Created end-to-end storyboards and interactive prototypes to challenge initial assumptions about linear AI generation.

The Core Strategic Shift
Moved from: “Where should we put AI?” ➔ to: “What should this product actually help the analyst accomplish?”
06 — The Product Architecture

Structuring by Context: Groups & Products

Enterprise datasets are too vast to dump into a single AI prompt. We structured the product model around bounded domains:

DOMAIN GROUP Loans & Credit Facilities
Product A: Corporate Lending (Dedicated glossary, schemas & mappings)
Product B: Retail Mortgages (Reuses loan domain knowledge)
The AI-Augmented Future Journey:
01. Start Context

Scope the AI to specific product domain.

02. Gather Data

Ingest schemas, sample records & specs.

03. Draft Glossary

AI creates initial shared business definitions.

04. Recommend

Suggest source-to-target field linkages.

05. Apply Expertise

Analyst reviews, corrects & refines.

06. Compound

Persist approved knowledge for next time.

07-08 — The Interaction Breakthrough

Designing for the Wait

AI processing is not instantaneous. If every generation step forced the user to wait, the product would devolve into a sequence of blocking spinner screens.

The Initial Linear Trap
Upload Files ➔ [BLOCKING WAIT]
Generate Glossary ➔ [BLOCKING WAIT]
Generate Mappings ➔ [BLOCKING WAIT]

Analyst is trapped on loading screens, breaking cognitive momentum.

The Asynchronous Solution
Dispatch AI Task ➔ Switch to another Product
Background Engine runs ➔ Live Status Badges
Task Completed ➔ Actionable Toast Notification

Analyst remains in continuous flow while AI works in the background.

Asynchronous Multi-Product Workspace Interactive State & Notification Simulation
Demo
Product: Corporate Loans READY FOR REVIEW

AI finished drafting 18 glossary definitions and 24 mapping recommendations in background.

Saved state: 2 mins ago
1. Multi-Product Switch jobs without losing progress.
2. Status Badges Processing, Ready, Needs Attention.
3. Leave & Return Persist state across sessions.
4. Live Alerts Notifications bring user back to result.
5. Parallel AI Multiple AI tasks run concurrently.
6. Stateful Memory Remember user adjustments.
09 — The Interaction Model

AI Assists. Humans Decide.

The core philosophy of the companion: AI reduces discovery friction; the analyst retains full accountability.

The AI Role
Engine
1. Suggest

Proposes glossaries and source-to-target field matches based on data sample patterns.

2. Explain

Provides transparent reasoning: schema lineage, field statistics, and semantic similarity scores.

3. Learn

Adapts when the analyst overrides a recommendation and updates domain memory.

👩🏽‍💻 The Human Role
Decider
1. Review

Inspects proposed business glossary and mapping candidates with context.

2. Validate & Refine

Corrects misclassifications, adjusts data types, and enriches business descriptions.

3. Approve

Formally signs off on integrations with full auditability and confidence.

10-11 — Craft & Leadership

Prototyping Under Ambiguity & My Contribution

Static screens were insufficient to evaluate asynchronous behaviors. I used Figma + Codex to build interactive prototypes that acted as thinking tools for leadership.

🧪 Key Behavioral Questions Answered Through Interactive Prototypes:
• What happens when AI is processing in the background?
• Where does the analyst go while waiting?
• How do they quickly return to unfinished work?
• What happens when a recommendation isn't trusted?
Leadership & Ownership Breakdown:
Me + PM Strategic
  • • Shaping the core problem
  • • Defining where AI fits
  • • Agreeing on future journey
  • • Product model structure
Me (Lead Craft) Execution
  • • Current journey mapping
  • • AI use-case exploration
  • • Storyboarding & flows
  • • Asynchronous interaction model
  • • Figma + Codex high-fi prototypes
Me + PM Principles
  • • Human/AI interaction rules
  • • Trust & explainability model
  • • Future roadmap scoping
🎯 The Critical Distinction: I wasn't handed a defined AI product and asked to decorate it. I actively shaped what the product should actually be.
12-13 — The Big Shift & Takeaways

The Big Design Shift

1. Context Matters

AI needs bounded scope (Domain Groups & Products) rather than raw enterprise noise.

2. Understanding First

The business glossary is part of the reasoning process, not a side feature.

3. Explainable Trust

Analysts require transparent rationale before approving suggested mappings.

4. Asynchronous Flow

Design around the wait so users can stay productive across multiple products.

5. Human Accountability

AI assists, suggests, and explains; the analyst remains the final decider.

6. Compounding Knowledge

Validated definitions become reusable enterprise assets rather than single-use outputs.

Portfolio Complementarity The Full Spectrum
01 · AFCS Financial Engine

Shows how I untangle massive existing legacy complexity, silos, and conflicting operational workflows.

02 · AI Data Companion

Shows how I lead 0-to-1 product definition, user modeling, and async interaction when nothing is defined yet.

“AI assists. Humans decide.”