Building a Tax Calculator for Estate* Lawyers
0-1 application
AI Workflow
Designing for AI
LawyerDoneDeal is a company building legal tech products that optimize workflows for lawyers and law clerks in Canada.
As the lead designer and product owner, I designed a document organization workflow for estate* lawyers.
* the type of law has been altered as a result of a NDA
TEAM:
1 product designer
1 developer
8 legal consultants
role
As the lead product designer and product manager on this project, I designed a tax calculator that helped lawyers calculate the amount of support their client would have to pay their ex spouse.
This meant understanding and defining the tax legislation and logic behind the calculations to describe in simpler terms its expected process and outcome.
This project is part of a larger 0→1 app that I joined during the proof of concept phase and currently working on refining functionality for launch.
projected IMPACT
Gaining Industry Traction
Canada's estate* legal software market has one dominant player charging over $200 per user per month, leaving smaller firms underserved. This fee is often disbursed to clients seeking legal justice.
At the time of filing, it is estimated around 64%-74% of clients are self-litigated. This calculator, offered as both part of the larger 0→1 app and as a standalone application will make it easier for underserved clients to access legal justice.
* the type of law has been altered as a result of a NDA
problem & user need
Lawyers assisting clients with their divorce affairs require the ability to calculate their client's income after federal and provincial taxes have been applied. However, this determination of net income requires an understanding of tax legislation that few estate lawyers have. Lawyers often trust and rely on software to assist them in these calculations. There is only one software in Canada's estate legal market, charging over $200 per user per month, leaving smaller firms who are unable to pay this fee, underserved.
key challenge
I don't have a tax legislative background, and there were few competitive products on the market.
To combat this, I encouraged my CEO to hire an intern who worked with an external tax auditor to build the backend calculation logic using Cursor AI. However, the build was updated almost daily, resulting in challenges documenting specs against a target that was moving faster than I could understand, scope, design, and spec for a developer.
approach
I utilized AI extensively to help speed my learning, spec writing, and design process.
Define
Dove deep into the legislation by understanding the law in plain language, flagging gaps in my understanding, and verifying AI’s answers with the source links.
OLD SCHOOL v1
Good ‘ol pencil and paper sketching work here to create some first iterations
scoping the build
Wrote a skills.md file for an agent to create a report every time I pulled new changes on VSCode. The report would indicate to me what changed, how this change can be explained by legislation and how it aligns with the MVP requirements.
Designing and Iterating
Refined early designs using Mobbin for inspiration, and Figma Make and Claude Design for further design iterations by prompting with UX theories, calculation logic, and what I wanted the calculator design to communicate.
user input interface design
I utilized a calculation output sheet offered by the competitor as a basis for my first iteration. Following Jakob's Law, and the idea that users expect a new product to work like ones they're used to, I mirrored the shape of the document on our software to reduce the learning curve.
The first iteration was visually mapped to the competitor's ouput sheet.
As I worked with the layout longer, and explored different variations on Figma Make and Claude Design, I ended up restructuring it so that the left side held all inputs and the right side held all outputs. This clearer separation allowed users to flip mental models when using the software - entering information on the left, and interpreting results on the right.
In FigmaMake's second variation, it shows an input vs. output layout, now implemented in the current design.
As the intern's build progressed, I created a skills.md file for an agent to create a report on the change, the legislation behind it, and how it aligns with the MVP requirements I had set out.
Once I understood the feature's purpose, I decided where it belonged: on the calculator screen, tucked away in additional settings for power users, or an administrator setting for anything that wasn't relevant to day-to-day use. Applying Hick's Law, this kept the primary interface with fewer visible choices resulting in faster decisions for the user.
feature
impact
QA effort
value proposition
implemented?
detailed tax input fields
low–med
high
Would allow lawyers to calculate support more accurately, but would require extensive tax legislation knowledge.
no
tax credits for child’s special expenses
med
low–med
Would allow lawyers to calculate support more accurately; requires low taxation knowledge as values would automatically populate.
yes, but hidden
Examples of features implemented by the Intern/Cursor AI and implementation decision based on QA effort and value proposition.
AI FORWARD INTERFACE DESIGN
The legal demographic is often known to be resistant to change. This is compounded by the high stakes of a profession where mistakes can be costly. As a result, AI-forward interfaces designed for this user group must preserve user agency by allowing users to revert to familiar input methods when needed.
Lawyers can upload documentation per line item. Each uploaded document has been mapped to the related court-ready folder.
next steps
This app is set to launch February 2027 and is currently being evaluated at the court and judicial level. At scale, this platform has the potential to meaningfully reduce the administrative overhead while providing smaller law firms another avenue to provide their clients with access to justice.