Our Work
Helping people make legal documents and get answers with confidence.
Rocket Copilot
Your AI Legal Assistant
I collaborated with the AI and Legal Innovation team at Rocket Lawyer to design Rocket Copilot, a legal AI assistant that helps users create documents and get legal answers confidently. The new experience improved user comprehension and increased document completion rates by 18%.
THEMES
User Interviews
Information Architecture
User Interface Design
Interaction Design
Prototyping
Usability Testing
TEAM
Me as the Lead Product Designer
Product Manager
Legal Engineer
Data Scientist
2 FE Engineers
2 ML Engineers
TOOLS
Figma
Lovable
User testing.com
Google Meet
TIMELINE
4 Months
The Problem Discovery
Creating legal documents can be confusing even with step-by-step forms. Users often:
-
Feel uncertain whether they’re filling the form correctly.
-
Don’t know what clauses mean or when to include/exclude them.
-
Drop off when they encounter unfamiliar legal terms.
We hypothesized that adding AI help could improve completion rate, confidence, and trust.
Research & Insights
These findings revealed a clear opportunity: users needed real-time, trustworthy easy explanations, not just static guidance.
User Interviews
8
37%
Small business owners and tenants described legal tasks as "nerve-wracking" and "full of guesswork"
Defining the Opportunity
How might we...
-
Help users understand complex legal language in real time?
-
Build trust in an AI assistant for sensitive legal contexts?
-
Reduce drop-offs without overwhelming users with AI chatter?
From this, I framed our design goal:
"Design an AI legal assistant that feels like a friendly, trustworthy legal guide, not just another chatbot"
Ideation & Exploration
Early Concepts
We explored multiple assistant forms:
-
Chat first interface:
conversational Q&A beside the document. -
Inline help:
AI suggestions embedded contextually next to each field. -
Hybrid approach:
quick inline clarifications + expandable chat for deeper help.
After testing wireframes, users preferred the hybrid model, calling it "non-intrusive" and "reassuring"
The Solution
Project Name
This is your Project description. Provide a brief summary to help visitors understand the context and background of your work. Click on "Edit Text" or double click on the text box to start.
Project Name
This is your Project description. Provide a brief summary to help visitors understand the context and background of your work. Click on "Edit Text" or double click on the text box to start.
Project Name
This is your Project description. Provide a brief summary to help visitors understand the context and background of your work. Click on "Edit Text" or double click on the text box to start.
Project Name
This is your Project description. Provide a brief summary to help visitors understand the context and background of your work. Click on "Edit Text" or double click on the text box to start.
Project Name
This is your Project description. Provide a brief summary to help visitors understand the context and background of your work. Click on "Edit Text" or double click on the text box to start.
Validation & Iteration
Usability Testing
We ran 3 rounds of usability tests (12 participants total) using a high-fidelity prototype.
Key findings:
-
92% said they trusted Rocket Copilot's explanations.
-
Time-to-complete document decreased by 23%.
-
75% described the assistant as "helpful but not overwhelming"
Design Adjustments
-
Added confidence labels (Based on legal best practices) to increase trust.
-
Improved the Copilot brand voice to feel more approachable, trustworthy, and aligned with Rocket Lawyer’s tone.
-
Fine tuned visual hierarchy to keep focus on the document, not the chat.
Launch & Impact
Post-Release Results
After release to 10% of users:
-
+18% increase in document completion rate.
-
-25% reduction in "need help" tickets
-
+2.1 average increase in user confidence rating (via post-task survey).
The success led to a full rollout across all legal document types, and eventually became the foundation for the company’s AI Copilot strategy.
My Contributions
End-to-End Design Leadership
Led end-to-end design from discovery through validation.
Prototyping & Experience Innovation
Created multi-modal prototypes integrating Figma + GPT sandbox environments.
AI Collaboration & Prompt Design
Partnered with legal engineers to translate domain expertise into AI prompts.
Data-Driven Impact Measurement
Collaborated with data science to define success metrics and dashboards post-launch.
My Contributions
What I learned:
Next Steps:
-
Building trust in AI requires transparency and restraint, not just accuracy.
-
Legal AI design is about balancing compliance, empathy, and clarity.
-
Small UX details (like phrasing, icons, or delay timing) deeply shape user trust.
-
Integrating user feedback loops to continuously improve responses.
-
Exploring how the assistant could proactively educate users beyond one document.