Written by Srishti Saigal

Project Overview

<aside> ✨

Everyone can create content; it can be created at scale, but how do we create quality content at scale?

That’s where UX Writers lead the process.

</aside>

More people are catching onto how AI can be leveraged in day-to-day activities. But not all users are sure how to go about it. In 2024, up and coming AI writing tools were an unfamiliar concept.

Users inevitably find themselves stumbling upon user interfaces where the use of AI is encouraged.

As a UX Writer, my challenge was to make users feel at ease using an AI writing tool in a resume-builder.

<aside> 🎯 Quick look at early user research sentiment on AI: In our previous research, users reported interest in using AI solutions that help them craft better content for their resumes. This works like Grammarly for resume making, or a solution that produces results from shorter phrases.

</aside>

Hypothesis: Users wanted to know how AI can help them create a presentable resume.

Goal: To understand which language was preferred. Was it the one that used AI promotional language (for ex: AI writing tool coming soon), or the one that did not explicitly mention AI, but used the tool in the background.

Problem: The problem was more of an opportunity. Why do we need an AI feature in our builder? ****While we had Text Tuner Content (TTCs) that was generated based on a user’s experience, it had some pros and cons. Our hypothesis was that AI implementation would help the user enhance this content (TTCs) further.

Pros of using Text Tuner Content (TTCs)

• Provided a good starting point (non blank pages)
• A good prompt for job responsibilities
• Provided pre-made content
• Delivered professional help
• Showed good writing style and action words
• job bullets aligned with the user’s job experience (most of the time)

Cons of using Text Tuner Content (TTCs)

• Too generic or too specific job bullets
• Irrelevant job bullets
• Not enough numbers or too many
• Descriptions that do not sound like the users: too eloquent, too many keywords, too complicated, etc.
• No quantifiable descriptions
• Not personalized enough descriptions, etc.

<aside> 💬 Step 1: AI language workshop

</aside>

<aside> 💡 Step 2: Competitive Analysis & Content Explorations

</aside>

<aside> 🤔 Step 3: Usability testing for UX writing

</aside>

<aside> 📌 Step 4: Learnings from UXR

</aside>

<aside> ✨ Step 5: Final look and results

</aside>

Step 1: AI language workshop

To better understand how to approach language for AI features, including instructional text, error states, and empty states, I organized a workshop for open discussion and brainstorming. Key stakeholders included: UX writers, product managers, designers, UX researchers, and program managers. We used FigJam to collect insights.