AI Governance | Sim Deol

Holistic AI

Designing Ethical AI at Scale – Case Study

SaaS B2B Web App User Onboarding

Introduction

An overview of the product design process.

Holistic AI is a pioneering risk management platform that helps enterprises and governments adopt AI responsibly. As industry leaders navigating the evolving regulatory landscape, Holistic AI empowers clients with the tools to ensure fairness, transparency, and compliance across their AI systems.

As the sole Product and UX/UI Designer at Holistic AI, a leading risk and compliance platform for responsible AI adoption, I led the design of three interconnected platforms: the AI Governance Platform, AI Tracker, and AI Safeguard. Each addressed a critical piece of the AI compliance puzzle: managing ethical standards, tracking global legislation, and ensuring safe LLM usage. Over two years, I helped shape a product ecosystem that not only kept pace with a rapidly evolving regulatory landscape but also empowered enterprise clients to navigate AI risk with clarity and confidence.

“Design isn’t just how AI looks, it’s how AI earns trust. Great UX makes the invisible understandable.”

The Challenge

Our goal was to deliver best-in-class AI risk and compliance tools to large enterprises, but we were doing so in a highly fragmented environment:

Each platform (Governance, Tracker, Safeguard) was designed independently on separate frameworks (Angular, React)

There was no shared design language or UI consistency

The industry itself was evolving quickly, and validating product decisions was often difficult due to a lack of precedent

For clients, this meant steep learning curves, manual onboarding processes, and inconsistent user experiences.

My Role

Sole designer across three core platforms

Led design system strategy and cross-platform unification

Acted as interim product manager: managed timelines, led scrum meetings, and facilitated cross-functional communication

Collaborated daily with engineers, data scientists, legal and policy teams

Conducted qualitative research with clients including Unilever, Mapfre, Starling Bank, and Mindbridge

The Tools

For collaboration, alignment and design.

The Strategy

Scalable and adaptive for a changing landscape.

I knew the long-term success of these products depended on a cohesive, modular design system that could bridge the gap between our disparate tools. I:

Spearheaded a transition from Angular to React + Tailwind for frontend consistency

Selected a component library compatible with the engineering team’s strengths

Created a unified design system that reduced friction and simplified cross-product development

Developed a future-state vision for a unified AI governance platform and presented it to leadership, realigning all teams under a shared north star

Design Process

Full oversight of the end-to-end development, from ideation to delivery.

My design process spanned the full product lifecycle, from research and ideation to high-fidelity prototyping and engineering handoff. I operated as both strategic thinker and executional lead, ensuring every design decision was grounded in user need, business context, and technical feasibility. This process allowed us to ship faster, reduce design–dev misalignment, and deliver polished, high-impact platforms that earned praise from both users and clients.

Discovery & Workshops: I facilitated workshops with internal stakeholders and enterprise clients to surface needs, uncover pain points, and define long-term product direction.

Competitive research: In an emerging and fast-evolving regulatory landscape, I conducted in-depth analysis of competitors and frameworks including the EU AI Act, NIST, and ISO, to ensure our products remained ahead of legislative change.

Design Systems Thinking: I partnered closely with engineers to select a design system that aligned with their frontend expertise, ensuring rapid adoption and seamless collaboration. We chose Shadcn components within a React framework, leveraging Tailwind CSS to accelerate development while maintaining consistency and scalability across products. I extended this setup by introducing design tokens for efficient scalability. While this approach enabled speed and cohesion, it did mean working within the aesthetic and interaction limits of the Shadcn UI framework, a trade-off that was necessary given the project’s timelines and team structure.

Prototyping & Testing: I created interactive prototypes (using Figma and Axure) to test complex workflows and edge cases with users and internal stakeholders.

Cross-functional alignment: I also led roadmap planning, facilitated weekly retrospectives, and ran daily stand-ups to ensure transparency, accountability, and continuous iteration across design, engineering, and policy teams.

Iterative QA & Refinement: I ran regular QA walkthroughs and post-release assessments to ensure fidelity between design and implementation and to rapidly address any usability issues.

Before

After

Ideation/Workshops

Research

User Journey/Flows

Development

The Products

Designing a Connected Ecosystem of AI Management Tools.

AI Governance

Centralised Ethical AI Management. The original platform was robust but rigid. I evolved it for enterprise scalability.

Interviewed Unilever to map their onboarding flow and discovered a pre-platform intake process

Proposed and designed a new intake flow that automated onboarding via file uploads and integrations with Confluence, GitHub

Collaborated with engineering and policy teams to embed lightweight pre-screening assessments

Result: significant decrease in manual effort and user frustration, helping retain Unilever as a long-term client and winning two new clients.

AI Tracker

Making Global AI Regulation Navigable. A real-time intelligence tool for monitoring AI legislation, policy, lawsuits, and more.

Created dynamic filtering, search, and visualisation tools to help users digest large, complex data

Balanced accessibility for legal and non-legal users alike

Designed scalable layouts to accommodate continuous data expansion

Collaborated with legal/policy teams to ensure design reflected nuanced legislative categories

Result: successful launch of a brand new product with over 1k sign-ups in one month.

AI Safeguard

Oversight for LLMs. A platform for monitoring and auditing the usage of Large Language Models (LLMs) within enterprises.

Designed dashboards for real-time monitoring and prompt testing against harmful behaviours (toxicity, hallucinations, PII leakage)

Created workflows for automated testing, risk flagging, and auditing

Ensured usability across compliance, data science, and engineering roles

Result: successful launch of a new product, with 2 major clients adopting the product.

Research and Validation

From Discovery to Direction

With limited resources, I leaned heavily on qualitative methods:

Conducted discovery and feedback workshops with clients like Mapfre, Unilever, and Starling Bank
Observed usage patterns to identify friction (e.g., repetitive onboarding, unclear risk flags)
Identified automation as a key value-add for large clients with high compliance overhead

These insights directly led to automation features that boosted platform usage and client satisfaction.

Outcome

Where Vision Meets Impact

60% increase in monthly AI system onboarding (due to automated intake)
Over 20k Sign-ups to the AI Tracker
Retained Unilever for another year despite prior cost concerns
Secured Mapfre as a new high-profile client
Delivered a roadmap and modular design system to unify three products into a future consolidated platform

Sim’s design leadership brought clarity to a fragmented product suite. We now have a unified vision and scalable UI across the board.

— Head of Engineering

First Ship, Then Shine

A rapid delivery mindset followed by a full visual evolution

A key consideration from the start of this project was the reality of limited engineering capacity. To move fast with a small team I selected Shadcn as our core component library. This allowed us to deliver a system that was scalable and fully aligned with Tailwind CSS and React Native. In the early stages the priority was function over visual refinement so that we could launch quickly and support rapid iteration.

Once the products gained traction and our technical foundations proved solid I championed a full visual upgrade. I revisited and refined every part of the interface to elevate the overall experience and bring the design closer to the vision we had for a market leading platform.

Reflection

Leading Through Complexity

What I’m most proud of is presenting a unified, strategic vision for a modular AI governance ecosystem, and seeing a once fragmented team, split across tech stacks and priorities, align behind it. This was more than design. It was leadership in ambiguity, foresight in a shifting industry, and diplomacy across disciplines.

Equally rewarding was the opportunity to co-create and drive two entirely new platforms, AI Tracker and AI Safeguard, from concept to launch. Both products emerged from cross-functional collaboration, user insight, and deep strategic thinking, and were met with strong client adoption. Watching those products go from whiteboard to enterprise impact was a powerful validation of my process, and of my ability to thrive at the edge of a fast moving frontier.