Overview
Atticus is an LLM‑powered contract assistant that helps non‑lawyers identify, understand, and negotiate critical clauses in minutes—without hiring a lawyer.
Challenge
In July 2023, my two co-founders built the first version in a week and launched on Hacker News. While hundreds of users tried it, only 1 user came back. They brought me on to fix it.
Goal
Redesign the experience to improve retention and unlock monetization.
Outcome
- Payments launched: $20 for 4 analyses
- 2.5% free-to-paid conversion
- Featured in Superhuman & Ben’s Bites
Research & Insights
I interviewed 10 users for 30 minutes each to understand their needs and pain points. These are the top problems I identified:
1
Buried Value
Users want a fast answer to: "Is this contract safe to sign?" This was hidden below the fold and hard to find.
2
Undiscoverable Chat
Users couldn't find or effectively use the Q&A functionality that enables deeper contract analysis.
3
Unclear Priority
Users didn't know what issues were urgent concerns versus nice-to-have changes to focus negotiations effectively.
Design Goals
To address our users' pain points and goal, I focused on:
Immediately Valuable
Surface actionable insights above the fold so users quickly experience the product's core value.
Intuitive Interaction
Redesign chat and follow-up experiences to be unmissable and feel like natural next steps.
Clear Hierarchy
Improve information architecture and add visual indicators to make output more actionable.
Solution
Using Figma, I prototyped a new UI in 2 days. Before implementation, I ran lightweight user testing sessions on the prototypes to validate whether users immediately recognized key outputs and understood what to do next.
Key Changes
Split-screen layout
Allows users to view AI analysis and chat side-by-side, encouraging deeper exploration and clarity without context-switching.
Prioritized issue list with explicit labels
Helps users focus by highlighting the most critical risks first and calling out what’s missing.
Smart chat suggestions
Gives users a starting point for questions—even when they’re unsure what to ask—making the AI feel more approachable.
Results
2.5%
Free-to-paid conversion
User testing gave us confidence to begin charging. Users had 1 free analysis and could purchase a 4-analysis pack for $20.
2
Industry Features
Picked up by tech newsletters Superhuman and Ben's Bites.
Key Learnings
Design for Knowledge Gaps
Non-lawyers don't know what to look for in contracts. Our design required clarity, prioritization, and surfacing what users didn't know to ask—matching how lawyers actually review documents.
Structure Enhances AI Quality
Instead of sending full contracts to the model, we determined contract type and party role first, then guided what to extract and analyze. This context-aware approach made the AI feel focused and trustworthy.
Focus on Goals, Not Trends
We didn't build a generic chatbot. We built an experience that delivered clarity, confidence, and clear next steps. The AI stayed in the background while insights and guidance took center stage.