Case Study Presentation
We Built the Intelligence. Nobody Knew It Existed.
Anuja Harsha — Lead Product Designer
DSML Hub · Cloud Software Group — WebFOCUS
We Built the Intelligence. Nobody Knew It Existed.
Cloud Software Group — WebFOCUS • Spring 2024 – Spring 2025
Problem
Low Adoption
Three AI features, invisible
Solution
3 → 1 Hub
Unified discovery surface
Timeline
Spring 2024 – Spring 2025
Architecture to production
Three Powerful AI Features. Near-Zero Adoption.
NLQ buried inside “Explore Data.” Insights in a submenu. ML in a data flow canvas. Millions invested. Nobody could find them.
Three AI features. Three different entry points. Three different mental models. Users didn't even know they existed.
My PM Karishma and I conceived the Hub together. My Director of Design and Head PM shaped the vision with us. Karishma pitched. VP approved. We owned it end-to-end.
The Fragmented Landscape
Path 1 — Finding NLQ or Insights
4 clicks · buried in a popup · side-by-side tabs
Path 2 — Finding Predict Data (ML)
4 clicks · completely different path · different mental model
Two separate paths. Two mental models.
No wonder nobody found them.
NLQ — buried in 'Explore Data' popup
Old Workflows
NLQ (Natural Language Query)
Automated Insights
ML Functions (Predict Data)
Upgrading the Building Blocks.
I already owned NLQ and ML. Insights landed in my lap — same engineering team, same PM, same QA. Before unifying them, I modernized each one individually.
NLQ Redesign
Suggested queries, conversational errors
Zero-intimidation entry point
AI-generated chart from natural language
Conversational errors — not stack traces
Insights Redesign
Auto-generated insight cards
Scannable card-based results
Architecture Before Tickets.
I pushed for things never done in the org. Karishma wrote tickets after seeing my mockups. 3–4 meetings a week across timezones — architecture already mapped before anyone asked.
The Solution — One Click
After — Everything Under One Roof
IQ Architecture
One Entry Point
Smart integration inside the existing Hub. No new infrastructure needed. Three scattered menus → one destination, one mental model.
IQ Plugin architecture — NLQ, Insights, and ML under one Hub
Dataset Selection Workflow
Unified dataset selection workflow
Hub Layout
How IQ fits into the existing Hub layout
Sketches & Early Concepts
Early concept sketching
Architecture wireframes
Responsive Design
First fully responsive Hub app
Four Iterations to the Final Hub.
V1 too dense. V2 too passive. V3 competed with Hub navigation. V4 — clean icon tiles — landed.
Design Iterations
IQ Hub concept: NLQ + Insights + ML in one place
V1 — Too Dense
First concept — too much information
V2 — Too Passive
Second iteration — too sparse
V3 — Competed with Hub Nav
Third iteration — competed with Hub navigation
V4 — Shipped ✔
The Navigation Fight
The lead architect — 20+ years — wanted list views. The WebFOCUS architect — 35+ years — backed him. My argument: lists are exactly why nobody found these features. Large tiles give immediate context.
The final IQ Discover page — shipped
Insights in IQ
Insights: Empty state
Insights: Data selected
Insights: Scannable card-based results
NLQ in IQ
NLQ: Ask questions in plain English
NLQ: Data selected state
NLQ: AI-generated chart
ML in IQ
Predict Data landing — inside IQ
Compare models in context
Model results — fitted values
Run model — explainability view
Run model — prediction results
Preview Data in IQ
Preview data — sample tab
Preview data — key analysis
Preview data — time series
IQ Walkthrough
The Room Full of Veterans.
Director of Engineering. Principal Data Scientist. Head PM. Lead Architect. All decades of tenure. I was 2 years in. Karishma and I were the only new people in the room.
Every Meeting, Outnumbered
Veterans with 20 to 35 years at WebFOCUS. My PM and I onboarded together. We pushed the vision together.
And They Were Listening
It hit me later — I was arguing about data flows and ML architecture with people who'd been here for decades. And they were listening.
Visibility Was the Solution.
NLQ redesign + Phi-3 model upgrade drove +25% adoption. Customers looked forward to NLQ the most. The Hub brings all of it — NLQ, Insights, ML — front and center.
NLQ adoption from discoverability alone. No feature changes. The highest-leverage design work: making existing features findable.
Scattered tools unified into one Hub. The patterns became the platform's AI strategy foundation. NLQ + Insights live. Hub implemented — final QA pending at departure.
What This Project Proved
We co-created this project as a team. The features didn't change. The visibility did. That's the highest-leverage design work there is.
“Anuja led UX design initiatives with remarkable creativity, empathy, and precision. She consistently demonstrated a deep understanding of user-centered design and the ability to translate complex product requirements into intuitive and visually engaging experiences. What truly stands out is her collaborative spirit and problem-solving mindset.”
— Karishma Khadge, Senior Product Manager
“Anuja made a significant impact modernizing UX across our legacy enterprise products. She brings a rare combination of strategic thinking, design intuition, and the ability to work seamlessly across product, engineering, and business teams. Anuja is bold in her ideas and consistently proactive in turning complex problems into practical, user-centered solutions.”
— Vijay Raman, VP of Product Management
When I returned from maternity leave, the engineering team shared that they valued the way I approached design — understanding the system, asking the right questions, making informed decisions. The experience reinforced something I already believed: context isn't transferable through documentation alone. It comes from genuine curiosity and investment in the problem.
Proper Tutorials and Onboarding
Tutorials & Onboarding: Initially proposed a dedicated "Get Started" panel. Through iteration, we improved the tutorials, Discover page, and Explore Data header instead — better UX, same goal.
Ecosystem Integration
Ecosystem Integration: Connect ReportCaster + IQ to auto-schedule generated insights. NLQ as a chat interface for the entire platform.