ULTIMARII
From 0 to 15 clients: Driving product discovery and UX in AI for regulatory work
PRODUCT DISCOVERY
PRODUCT STRATEGY
AI UX
INFORMATION ARCHITECTURE
LLM DESIGN
REGULATORY PROFESSIONALS SPEND hundreds of hours manually reviewing legal and compliance-heavy documents. A process prone to inefficiencies and human error.
GOALS
Ultimarii set out to build an AI-powered research platform that could:
Centralize large document repositories, making structured and unstructured data instantly searchable
Leverage AI-driven summarization to extract key insights in minutes instead of hours
Enable users to refine queries and interpret AI-generated insights while maintaining transparency and accuracy
The challenge wasn’t just designing a research tool. It was also about defining how users interact with an LLM (Large Language Model) in a high-stakes regulatory setting where every insight needed to be verifiable.
01
Sprint 1: Laying out the groundwork
Our first sprint focused on defining core user workflows and interactions with AI. I developed a detailed storyboard that mapped out:
AI-powered search and retrieval, allowing users to input regulatory queries
Smart document summarization, breaking down lengthy reports into structured insights
AI-assisted workflows, where the system would suggest key documents, highlight critical clauses, and assist in report drafting
IMPACT
This storyboard became a powerful pitch tool for Ultimarii’s early enterprise clients, helping them visualize how the AI-powered system could transform their workflows.
tl;dr
THE PROBLEM
Regulatory professionals struggle with navigating dense, compliance-heavy documents. Existing research workflows were time-consuming, fragmented, and highly manual.
Ultimarii aimed to develop an AI-powered system that could centralize data, speed up research, and provide AI-assisted insights with source traceability.
MY APPROACH
Led product discovery and UX research to define key user needs and interactions with AI
Mapped out emerging LLM user behaviors, refining AI-assisted search and document analysis workflows
Designed a scalable information architecture and UI framework to support AI-driven regulatory research
THE RESULT
Helped Ultimarii secure 15 enterprise clients within its first year
Delivered an intuitive experience for handling complex document workflows
Established a UX/UI foundation that accelerated product development and scalability
Walks through of the user flows of two core MVP goals: (1) Document Repository Searching and (2) Document Summarization, Analysis, and Inspection.
Overview of the low fidelity wires for flow 1 and 2. I added notes to myself for potential branching of user flow, edge cases that needs to be considered, and potential interactions to explore.
02
Sprint 2: Designing for large-scale data and LLM interactions
With the product concept validated, the next sprint focused on simplifying large data interactions and refining the AI-powered user experience.
Key UX challenges included:
❗️
Handling massive datasets - how do we structure large volumes of legal text into digestible insights?
❗️
Ensuring AI-generated summaries are accurate and trustworthy
❗️
Designing intuitive LLM interactions - how do users engage with AI beyond simple queries?
SOLUTION
I tackled these challenges by:
Refining AI-powered search to provide contextual rather than keyword-based results
Designing a structured document viewer that allowed users to see source-backed AI summaries
Mapping emerging query interactions like how users input, refine, and validate AI-generated insights
The goal was to create a balance between automation and user control, ensuring professionals could trust the AI without feeling like they were working with a black box.
Sprint 2 unpacked current proven way to visualize large data sets, defining emerging types of query interactions unique to LLM platforms, and mapping out the UX and UI for what those interactions can look like.
03
Sprint 3: Iterating on emerging complex user needs
As we approached MVP readiness, the final sprint focused on optimizing navigation, workflows, and usability. We tackled:
Nested document navigation, ensuring users could manage multiple regulatory files without friction
Interactive AI responses, refining how the system guided users in refining their queries
Scalable UI patterns, ensuring the product could evolve with growing complexity
By the end of this sprint, the AI-driven research tool was an enterprise-ready product.
Annotated view of the proposed revised interface
Overview of the main interface mapping out the various states and key library interactions
Overview of the full sprint 3 with a breakdown of the main interface functionality and a breakdown of core feature functionality for the Collections section and chat interactions
WHAT MY COLLEAGUES SAY
Trace is incredibly dedicated and often goes above and beyond to make deliverables or create the best product. She is a very strong leader and has immense skill in leading client workshops, especially in the UX and CDR space.
04
Impact on Ultimarii's growth
Ultimarii launched its AI-powered platform with 15 enterprise clients, including major energy and utility companies, within its first year. The UX work directly contributed to:
IMPACT
The UX work directly contributed to:
Increased adoption - Clients found the interface approachable and easy to integrate into existing workflows
Faster research cycles - Professionals could now retrieve insights in minutes instead of days
Securing $2M in funding, with a strategic round planned for 2025
High fidelity interface design of the landing page of the web platform from Sprint 2. Actual MVP interface is currently available under early access only.
05
What I learned
When I first started working with Ultimarii, I didn’t fully grasp the scale of its ambition. What started as a UX challenge quickly became a systems thinking problem—how do you create an AI-powered interface that not only processes vast amounts of data but also makes that intelligence accessible and actionable for users?
AI UX is still an emerging field - We had to define user interactions with an LLM without established best practices
Trust in AI is a design problem - Users needed confidence that AI-generated insights were accurate and transparent
Strong UX strategy can shift product direction - By deeply understanding user needs, we influenced how the AI system itself was structured
Initially, there was skepticism about whether UX could meaningfully impact Ultimarii’s AI development. By the end of the sprint cycles, the production team relied on our UX frameworks as a core part of their process.
This project reinforced my passion for solving complex design challenges, particularly in AI-driven products. Seeing how the work helped shape a tool that is actively changing how regulatory research is done at scale was an incredible validation of UX’s role in AI innovation.
FINAL THOUGHTS
Ultimarii is now positioned as a leader in AI-driven regulatory research, and the work done in these sprint cycles laid the foundation for its continued growth.
SHOUTOUTS TO—
Chantelle Little - For trusting me with the opportunity to fly solo on the initial sprint, keep the momentum going and in building a great collaborative partnership with the Ultimarii team.
Josh Malate and Doug Schweitzer - For the easy collaboration, the free flowing whiteboarding sessions, for sharing knowledge about regulatory work and for the trust to get Ultimarii into MVP
Duane Wood and team - Also for the trust! Eventually. I kid. :p Truly thankful for the wonderful collaboration, the deep conversations unpacking the complexity of the design challenges, the grace and patience as we evolved the product into what it is today.
