Overview

Name: Elisa Fang

Role: Product Designer (Solo)

Timeline: 3 weeks design, 3 weeks engineering

Company: Supio (AI Legal Tech Startup)

Impact: 90%+ firm adoption, 30% operational efficiency gain

Longer prompt selected.png


Context

Supio is an AI legal tech startup focused on improving workflows for personal injury attorneys and paralegals. Our customers span a wide spectrum of technical sophistication—some firms eagerly adopt cutting-edge AI technology, while others harbor deep concerns about AI risks and reliability.

As AI technology was still new to the legal industry, most customers weren't familiar with effective prompting practices. They didn't know how to craft prompts that would yield useful results, or how to iterate when initial attempts fell short.


The Problem

Even though our model improves, engagement and customer satisfaction wasn’t improving.

While our engineering team continuously improved our AI models, we weren't seeing corresponding increases in chat engagement. This disconnect led us to investigate: was the barrier our technology, or the user experience?

The answer came from our Customer Experience team, who flagged patterns in our customer forums. Our most engaged users were sharing prompts that "worked really well" for specific legal tasks—and some firms were even maintaining Word documents of effective prompts that they passed around their offices.

The Root Cause: Ineffective Prompting

User complaints often stemmed not from AI limitations, but from poor prompting habits and a lack of understanding of how to approach more complex questions. We identified two critical patterns: