Designed an AI-powered resume generation tool reducing time-to-apply by 60%.
Context
Workruit's data showed a stark pattern: candidates who completed their resume on the platform were 4× more likely to get hired. But only 38% of registered candidates ever finished one.
The existing resume builder was a form. Forty-seven fields, zero guidance, unlimited frustration.
The Opportunity
With LLMs becoming viable in early 2023, the product team asked: what if the AI did the heavy lifting?
My job wasn't to design a chatbot. It was to figure out how much of the resume creation experience could be offloaded to the AI without making candidates feel like they'd lost authorship of their own story.
Design Thinking
The key tension: AI suggestions feel helpful when they save time, and creepy when they feel presumptuous. The line is thin.
I designed a model where candidates provided four inputs — job title, years of experience, industry, and two or three bullet points about their work — and the AI generated a complete first draft. Every AI-generated section was visually marked, editable inline, and came with a "regenerate" option.
This kept candidates in control while dramatically reducing blank-page paralysis.
Results
Time-to-apply dropped 60%. Application completion rate jumped 45%. The feature launched to a 4.8-star rating on the app store with over 50,000 reviews in the first quarter.