95% User Satisfaction Through Agentic AI

Led UX for Thomson Reuters’ first agentic AI tax advisory product, adopted by hundreds of companies in its first month with 95% user satisfaction.

95% User Satisfaction Through Agentic AI

Led UX for Thomson Reuters’ first agentic AI tax advisory product, adopted by hundreds of companies in its first month with 95% user satisfaction.

CLIENT

Thompson Reuters

CLIENT

Thompson Reuters

Role

Lead UX Designer

Role

Lead UX Designer

Service

Enterprise SaaS, Taxtech

Service

Enterprise SaaS, Taxtech

Yellow Flower
Yellow Flower
Accountant Prototype
Client Portal Prototype

Context

Context

Thomson Reuters identified a +$100M opportunity to revolutionize the tax advisory space with an AI-powered solution. The goal was to help firms enhance their advisory offerings and generate data-driven strategies.

As a Lead UX Designer, I was selected to join a specialized team that traveled to the Minnesota office to collaborate with cross-functional company leaders. Our mission was to brainstorm, design, and bring a functional prototype to life in time for a live demo presented to the SYNERGY 2024 audience.

Photo by Thompson Reuters

Problem & Opportunity

Problem & Opportunity

Challenges in Advisory Services

Advisory services require deep expertise, making it difficult for firms to scale their offerings efficiently. The gap between experienced professionals, who possess deep advisory knowledge, and junior accountants, who need structured guidance, was a major pain point.

The Role of AI

By leveraging AI, I aimed to simplify the advisory process, enabling firms to:

  • Add new clients and analyze financial data seamlessly.

  • Identify advisory opportunities through AI-powered recommendations.

  • Provide step-by-step strategic execution for CPAs.

Strategy & Process

Strategy & Process

Cross-Functional Collaboration

The project involved close collaboration with business strategists, AI engineers, researchers, design directors and specialized entrepreneurs. Our workshops in the Minnesota Thomson Reuters offices focused on aligning business objectives with user needs.

Competitive Insights

During the project, I conducted a rapid competitive analysis on-site, assessing existing solutions and identifying gaps where our AI-driven approach could differentiate itself:

  • Existing solutions rely heavily on manual data entry, limiting efficiency and scalability.

  • Most platforms are designed for users with significant advisory experience, making them inaccessible to junior professionals.

  • Current products support advisory as a feature rather than a core function. They are not built with an "advisory-first" mindset.

Research & Ideation

To understand the workflow of advisory professionals, we, as a team, conducted:

10 pre-Synergy and 56 Synergy user interviews with tax firm owners, partners, directors, CPAs.
Ideation sessions to map out user journeys and interaction flows
A deep dive into how AI could enhance decision-making and support the jobs-to-be-done in the advisory process

From Vision to Screens

From Vision to Screens

Designing to Drive Decision-Making

Initial Wireframes

After days of insightful but indecisive conversations with directors, I knew we needed something real to react to, something to move us from “what if” to “let’s build this”. So I did what I always do when things feel abstract: I opened up my design tool, threw down some rectangles and boxes, and started shaping what I could see in my head.

I teamed up with two brilliant principal designers, and we began bouncing ideas, challenging each other, and building on sparks of inspiration. Those sessions lit a fire in me. I became obsessed with getting the full flow out of my head and onto the screen, deep into two focused days, designing every screen that could bring this AI advisory solution to life.

But this wasn’t just about UI. I imagined an experience that could scale, grow, and actually think with the user. A system that offers guidance at every step, whether through a wizard, a smart loader, an automated email, or surfacing the right opportunity at the right time. A truly agentic solution.

I poured those ideas into multiple mockups, always anchoring back to what I knew about our users. And then I waited for feedback from stakeholders, client sessions, anyone who could tell me if we were on the right track. It was the kind of leap that gets real decisions made.

Refining the Experience & Building the Prototype

Not everything I designed solved the right problem at first. Some screens looked polished but missed the mark, and I had to go through tough iterations, defending some decisions, letting others go. That’s the beauty of working with smart, open collaborators: the space to challenge each other without ego.

The TR design system helped us move quickly to high-fidelity, but it didn’t cover everything this experience needed. We had to go beyond the library, crafting custom interactions that helped users discover information naturally and move through advisory steps without breaking focus.

In the end, the prototype wasn’t just a visual tool, it was a working model of how the product could think and guide. Something real we could test, show, and use to drive decisions.

The prototype focused on 3 core functionalities

  1. Client Data Input & Analysis: A seamless onboarding experience for entering and analyzing financial data.
  1. AI-Generated Advisory Strategies: Intelligent recommendations tailored to client needs.
  1. Strategy Execution Guide: A structured, guided approach to implementing advisory solutions.

Insights & what's next

Insights & what's next

Takeaways

The demo at SYNERGY 2024 provided valuable real-time feedback from industry professionals. Some key takeaways:

The praise
  • Users believed the insights and strategies can scale and be shared firm wide.

  • Users expressed that the solution saves time on tax research and report generation, making the advisory process more efficient and less reliant on manual input.

  • Users found that this solution provides a streamlined way to create and deliver comprehensive reports to advisory clients, which adds a significant value.

  • Users believed that the solution has potential to get junior staff involved, addressing staffing shortages and enable knowledge transfer from senior to junior members of the firm.

The challenges
  • The cost (mentioned by 10 participants) might be a significant barrier for small firms.

  • Data privacy and AI accuracy (8 mentions). Concerned about the data storage and the potential that AI might generate misleading or inaccurate information.

  • Change management and training (7 mentions), the need for effective integration into existing workflows, along with work needed for staff training and stakeholder buy-in.

Photo By Thompson Reuters

Impact & Future Steps

The success of this project positioned AI-driven advisory solutions as a core focus for Thomson Reuters. The key impacts included:

Reflections & Lessons Learned

Leading the UX design on this high-stakes, fast-moving project taught me more than just how to ship a compelling prototype. It reshaped how I think about building for complexity, scale, and real impact.

Cross-functional collaboration is everything in AI-driven solutions

This project wouldn’t have been possible without constant alignment between design, product, engineering, UX researchers, and business directors. AI introduces nuance and ambiguity... what’s possible, what’s ethical, what’s useful. Being in the room with experts from different disciplines allowed us to shape a solution that was both technically feasible and strategically sound. UX acted as the translator between user needs and AI capabilities.

User-centered design becomes even more critical in complex financial workflows.


It’s easy to get lost in the jargon, the regulations, and the data-heavy nature of tax advisory. But behind it all are real people, junior accountants trying to build confidence, experienced CPAs looking to scale their value, firms striving to evolve. Every decision we made came back to: “Does this reduce friction? Does this support the user’s thinking? Does it help them make smarter moves, faster?”

AI product development is iterative, and humbling.


We didn't get everything right the first time. Some ideas looked promising on paper but didn’t resonate in testing. Others emerged through feedback and collaboration. Real-world input, especially from the SYNERGY audience, was invaluable. It reminded me that when you’re designing with AI, you’re not just building interfaces; you’re shaping decision-making environments. That takes time, trial, and a lot of learning along the way.