Context
LexFriend is developed as an AI-supported workspace that helps legal professionals structure cases, draft documents, and prioritise actions. Enabling them to maintain control over their work and spend less time on repetitive tasks.
CHALLENGE
How might we help legal professionals regain control over complex dossiers by reducing fragmentation and administrative overhead, without compromising legal accuracy or autonomy?
MY ROLE
Lead UI/UX Designer responsible for end-to-end product design.
Faced with a fragmented product and no shared foundation. I made a deliberate call: fix the underlying system before layering features on top.
That meant building a design system in parallel with feature delivery, and designing AI workflows that worked for lawyers and secretaries.
TIMELINE
Feb 2025 - Feb 2026
CLIENT
LexFriend
Context
LexFriend is developed as an AI-supported workspace that helps legal professionals structure cases, draft documents, and prioritise actions. Enabling them to maintain control over their work and spend less time on repetitive tasks.
CHALLENGE
How might we help legal professionals regain control over complex dossiers by reducing fragmentation and administrative overhead, without compromising legal accuracy or autonomy?
MY ROLE
Lead UI/UX Designer responsible for end-to-end product design.
Faced with a fragmented product and no shared foundation. I made a deliberate call: fix the underlying system before layering features on top.
That meant building a design system in parallel with feature delivery, and designing AI workflows that worked for lawyers and secretaries.
TIMELINE
Feb 2025 - Feb 2026
CLIENT
LexFriend
Context
LexFriend is developed as an AI-supported workspace that helps legal professionals structure cases, draft documents, and prioritise actions. Enabling them to maintain control over their work and spend less time on repetitive tasks.
CHALLENGE
How might we help legal professionals regain control over complex dossiers by reducing fragmentation and administrative overhead, without compromising legal accuracy or autonomy?
MY ROLE
Lead UI/UX Designer responsible for end-to-end product design.
Faced with a fragmented product and no shared foundation. I made a deliberate call: fix the underlying system before layering features on top.
That meant building a design system in parallel with feature delivery, and designing AI workflows that worked for lawyers and secretaries.
TIMELINE
Feb 2025 - Feb 2026
CLIENT
LexFriend
Discovery
After interviewing lawyers and secretaries at 2 small (10-20 employee) law firms, one thing became clear: the problem wasn't missing software. It was the manual labor that costed a lot of time. Lawyers had to do a lot of time consuming administration and secretaries couldn't act without interrupting a lawyer for context. Both were working around a tool that wasn't specified to their needs. That became the lens for everything I designed.
Starting with the foundation
In my first weeks, I found a fragmented codebase: inconsistent components, hardcoded spacing and no shared UI foundation. This wasn't just a design issue, it was generating rework and implementation guesswork across every sprint. I identified this early and made the decision to start with the foundation before adding new workflows on top.
Before having a foundation.
Before having a foundation.
After having a foundation
After having a foundation
Building the system
I was the only designer in a team of four developers. Hence why I built the system alongside feature delivery. Every component I added was one I was already designing that week. I documented patterns as they surfaced in real work and introduced the token structure incrementally. This kept the system grounded in actual product needs and didn't slow the team down. It also meant the system was in use from day one and not a parallel project that lived in Figma that never got shipped.
Step-by-step proces
01 / 03
Defining the principles
The team had no shared design language. Every screen was built slightly differently. Not because of bad decisions, but because nothing was agreed on. I defined a set of principles to create that agreement: consistency over creativity, components first, and nothing custom unless nothing existing fits. Small rules, but they prevented a lot of drift.
Applying the system to product workflows
With the system in place, I applied it to the workflows that mattered most: time registration and dossier management. Both involved AI-assisted actions.Introducing AI into their daily workflow wasn't just a UI challenge. It meant designing interactions that didn't require users to understand how the AI worked, only what to do next. The learning curve had to be as low as possible.
Highlighted workflows
01. Time registration
PROBLEM
Time registration was fragmented, manual, and often done after the fact. This led to lost or inaccurate billable hours, and low trust in the data.
TRADE-OFF
Letting AI interpret notes introduces a small learning gap on how to interact with the AI assistant.
SOLUTION
By introducing AI-driven time registration the manual administrative effort dropped, and legal professionals gained a reliable foundation for billing and reporting.
02. Dossier summary
PROBLEM
Lawyers and secretaries worked from different information. Secretaries lacked case context, which meant constant interruptions for lawyers to explain status, priorities, or financials.
TRADE-OFF
An AI-summary is only as good as the data in the dossier. If logged hours, tasks, or documents are incomplete, the summary misleads rather than helps. Trust in the output requires discipline in input.
SOLUTION
SOLUTION
A generated summary on the dossier level gives both roles the same picture at a glance (case status, outstanding financials and suggested next actions).
Impact
Because the product was in active development during the project timeline, there weren't post-launch metrics. The impact indicators are qualitative.
Because the product was in active development during the project timeline, there weren't post-launch metrics. The impact indicators are qualitative.
REDUCING LEGAL WORK'S ADMINISTRATIVE BURDEN
TIME REGISTRATION
Lawyers stopped losing billable hours to manual registration. Time was logged when it happened, not guessed at the end of the day or month.
CROSS-ROLE ALIGNMENT
The information gap between lawyers and secretaries was closed. They had one single source of truth they both have direct access to.
FOUNDATION FOR SCALE
Feature delivery stopped generating new inconsistencies. Every new workflow I designed could be built from existing components, without alignment rounds or rework.
FOUNDATION FOR SCALE
Feature delivery stopped generating new inconsistencies. Every new workflow I designed could be built from existing components, without alignment rounds or rework.
Reflection
Designing AI into complex, legal specific software taught me where AI actually helps and where it creates new problems. The margin for error is different when your users are lawyers billing clients. That changes how you design confirmation states, how much you expose the AI's reasoning, and how much control you leave with the user.
Next case
© 2026 Marcel. M - All rights reserved.