Designing an AI-Ready Enterprise Data Platform
Overview
This project focuses on the redesign of an enterprise Data & AI Marketplace, a platform used by over 50K employees ( 300K after the redesign) to make enterprise data accessible, searchable, and actionable across global teams. The goal was to modernize the legacy experience while integrating AI capabilities, making it scalable, role-aware, and ready for the next generation of intelligent data use.
Organization
Accenture Song
Role
UX Designer
Duration
6 months
Team
2 UX Designer + 1 Design Manager
What I Did
Redesigned core workflows and interfaces to improve clarity, scalability, and user efficiency.
Collaborated with cross-functionally with developers, product managers, and global design teams to align design, technical feasibility, and business goals.
Designed AI-ready interactions and personalization systems to enable adaptive, intelligent experiences.
Final Design
Redesigned Marketplace at a glance
250+ Frames Redesigned
100+ DS components
Chat Interface
15+ Forms and Wizards
Data Viz Dashboards
Jump to Redesign
IMPACT
%
Reduction in manual errors through automation.
Problem
A rigid, outdated platform that couldn’t scale with growing data needs or AI capabilities
Users struggled to complete key tasks as the platform grew cluttered, rigid, and impersonal.
Rising data volumes and AI workflows exposed scalability issues, slowing teams and decision-making speed.
Let's Dive Deep!
Background
What is a data marketplace?
A centralized data platform used across global teams to find, publish, and govern enterprise data
10k Monthly Active Users
300K Users
15 Countries
Key Actions supported by data marketplace
Supports business teams like Finance, HR, Sales, Marketing, and IT
Helps leaders at global and regional levels make data-driven decisions
Used by roles such as financial managers, analysts, and operations leads
Makes big data accessible for tracking costs, profits, and performance across teams
Persona
Designing for High-Stakes Users Across Business Functions
Our users held dual roles, both as stakeholders setting strategic direction and as active users relying on the platform for daily decision-making. Our users spanned product leaders and decision-makers across critical domains. This meant the platform had to be clear, reliable, and deliver real business value.
Research
Conducting Research in Stages
The scale and complexity of the platform brought challenges that went beyond usability. Many of our users were also product leaders themselves, making it even more important that every design choice was informed, defensible, and grounded in evidence.
Heuristics and Accessibility Issues that surfaced during initial research.
Goals
Balancing User and Stakeholder Goals
To guide the redesign, I compared what users needed with what stakeholders expected, finding overlaps and trade-offs that shaped a direction balancing usability for users with scalability and impact for the business.
User Pain-Points
↳ Information overload
↳ Inconsistent UI elements
↳ Inefficient Navigation
↳ Unclear data hierarchy
↳ Inconsistent UI elements
Me
Business Goals
↳ Integrate AI
↳ User Adoption + Satisfaction
↳ Improve Data Accessibility
↳ Scalable systems
↳ Implementation feasibility
Strategy
Transforming a legacy data platform into an AI-ready experience
As the sole designer on the MVP, I quickly saw how deep the structural gaps ran. I advocated for a phased strategy to solve urgent usability issues first, rebuild what was broken, and then design with scalability and AI-readiness in mind.
Phase 1: MVP & Iterative Discovery
Targeted improvements and new features delivered while learning from rapid testing and real-time feedback
Phase 2: Rebuilding the Foundation
Addressing structural issues through redesigned flows, components, and systems
Phase 3: Integrating AI
Creating a platform ready to support evolving GenAI use cases
Iterative Discovery
Using the MVP as a Sandbox for Discovery
As the sole UX designer for the MVP, I led all design sprints for 4 months in close partnership with PMs and developers, rapidly prototyping, testing, and iterating on core screens.
This phase became my sandbox for discovery: validating ideas, uncovering workflow gaps, and translating learnings into fast-impact features that shaped the product’s direction. While addressing immediate pain points, I also built the foundation and insights that informed the larger redesign in Phase 2.
A snapshot of how new ideas flowed through daily collaboration, aligning goals with PMs, evaluating feasibility with developers, and translating discoveries into tangible MVP releases.
Even as the youngest member of the team, working on the MVP gave me the most ground-level knowledge, positioning me to advocate for users and influence decisions beyond my scope.
Now Lets' Dive Into How I Made Support More Personal Within MVP Limits
“
Can we explore a way to make support more self-service?
- Request from PM
Proposing a Role-Based Chat
I proposed a chat to bridge the gap before the redesign - personalizing answers by user type, easing navigation, and keeping engagement high.
Balanced personalization and scalability
Adapted to three key user roles
Reduced clutter through guided flows
The result was a smoother, self-service experience that boosted CSAT by 25%
While a static support model might have offered a quick fix, I saw the chat assistant as a scalable foundation. It balanced immediate usability with the long-term vision for a more personalized platform.
Re-design
Redesigning key flows, core screens, and system foundations for clarity, usability, and scale
Redesigned Core Screens
Landing Page, Product Details, Search Results
Introduce Role-Aware Flows
My Assets, My Profile, Notifications, Subscription, Wizards
Core Screen
Starting with a homepage that gave users everything except what they needed.
“
Honestly, I skip most of this. I don’t even know what I’m looking at sometimes.
- P15, Daily User, Stakeholder Interview
The legacy homepage was static, cluttered, and ignored user intent. No differentiation for roles, no clear CTAs, and irrelevant cards meant users often hit dead ends.
Rebuilding the Landing Experience: Product Card
A redesigned product card layout to surface trust signals, key metadata, and dataset context at a glance. The original card layout included elements that didn’t support decision-making, such as irrelevant icons, unused review stars, and inconsistent CTA colors. Key information like access status was missing, and the design created visual clutter, making it harder for users to quickly scan and act.
As part of the core user experience redesign, the card was restructured to prioritize clarity and utility. Unnecessary elements were removed, the CTA color was standardized to purple, and space was optimized by limiting title length. New features like access status, bookmarks, and notifications were introduced to support personalization and improve user efficiency.
A role-aware, personalized homepage to surface relevant insights and drive faster decisions
I restructured the homepage to prioritize action. CTAs were clarified, irrelevant modules removed, and sections tailored by role. The result: less noise, more relevance, and faster access to key data.
Role-Aware Flows
Introducing Much-Needed Personalization: My Assets
A dedicated space for users to manage published and favorited assets in one place.
“
I use the same datasets every week, but I have to search for them every single time.
- P3, Daily User, Stakeholder Interview
Having a platform where users accessed the same data daily but had to re-search it from scratch revealed a major usability gap. The absence of a dedicated space for recurring tasks slowed users down and created unnecessary frustration.
I introduced the My Assets page, a centralized place to manage published and favorited data products. It reduced search reliance and cut access time significantly.
Integrating AI
Designing for Smarter Discovery Before AI Infrastructure Was Ready
We couldn’t wait for AI to catch up. The existing platform was losing users, and backend upgrades would take time. So, we designed scalable AI-ready features in phases. These upgrades ensured the platform stayed relevant, even as tech and expectations continued to evolve.
Introducing natural language search to improve dataset discoverability and reduce friction
Legacy: Search required exact dataset names, making discovery rigid and prone to failure.
Redesigned: Added tag-enhanced filters and helper text to reduce failed queries and surface metadata.
Future State: Designed a GenAI assistant that supports conversational queries and task automation, like requesting access or exploring usage insights directly from the search bar.
Dynamic homepage cards tailored by role, behavior, and usage to surface what matters most
Legacy: Static “popular this week” cards showed generalized content irrelevant to individual roles.
Redesigned: Role-based collections introduced relevance by department (e.g., Finance).
Future State: AI-driven cards tailored by role, past usage, company events, and search behavior, providing dynamic, high-impact content on login.
Design System
Standardizing components and patterns to ensure consistency and accelerate future development
Why It Was Needed
Teams across the Data & AI Studio were using inconsistent design systems, leading to fragmented UI, accessibility issues, and inefficient handoffs. An audit revealed problems like misaligned padding, color mismatches, and complex variants.
What I Designed
Collaborated with global teams to create a unified, responsive design system that supported diverse product needs while ensuring accessibility and consistency. This became the foundation for the Accenture Design System Playbook.
Colors
Typography
Icons
Accordions
Buttons
Filters
Cards
Tables
Alerts
KPIs
Navigation
Charts
More
Takeaway
Less is More - in enterprise UX, the real challenge isn't designing new components, but ruthlessly simplifying existing ones. While stakeholders will always demand new features, our real value comes from designing less: removing the unnecessary, consolidating the redundant, and fighting for only what truly matters to users.
Cross-functional collaboration was key - Aligning designers, developers, and stakeholders ensured solutions were both usable and scalable.
Stakeholders speak different languages - Learned to translate UX findings into business metrics that resonated with different leaderships.
Hear From My Team
More
Due to NDA restrictions, the visuals from my work at Accenture are limited and masked. Feel free to reach out for more details.
Up Next
Design System
Enterprise
B2C
Sponsored Project
UI Enhancements
