Content Design > AI Term Research Assistant

Lexi, your competitor terminology research assistant

Overview

Product overview
- Powered by AI, Lexi scans thousands of online resources, performs deep research, and delivers clear competitor summaries and term recommendations in minutes.
- What used to take around two weeks of UX writing research for a major product now happens almost instantly.
- Lexi not only helps me uncover more competitors and terminology than I could on my own, but it’s also becoming part of 170+ writers’ and other product stakeholders’ workflows—for inspiration, validation, and major time savings.

How it works
- Lexi performs in-depth competitor terminology research to help writers discover the best terms for their Wix products
- Input: Share your product/feature name + term researching.
- Output: In minutes ,you’ll receive In depth analysis of top competitors, most used terms, and the recommended term for your Wix product.

Stakeholders
- Writers Guild at Wix (170+ writers), Head of Product in my domain (Wix Payments), Hackathon Team of 4

Challenges
- LLM model and prompt issues: Initially the LLM models were hallucinating by repeating themselves and providing generic responses. Reiterating the prompt 10+ times and choosing deep research models solved the problem.
- Limited time: During the hackathon, my team had 2 days to build the concept, app and presentation. Then we only had 3 minutes to present it.
- Uncertainty about where to host the product: There were issues connecting the LLMs API keys to Base 44 code-less app, custom Wix Chat wasn’t a deep research model, Gemini Gems were not shareable, etc.

Process break down

1. Create concept over one month

I knew I wanted to make the Content Design research process more efficient, but it took some time to narrow down my ideas and pinpoint the best place to focus to get results and to provide a quick win for writers and other stakeholders across varying products at Wix.

2. Recruit three team members

In utliized our writer’s two day AI Hackathon to recruit a team to help me bring this concept to life. I chose writers from different areas of expertise; localization, AI custom chat expert, and knowledge base.

3. Craft and reiterate prompt

I wrote, reiterated, and tested a strong prompt. This was the brain of the entire product.
I asked each LLM I tested to analyze UX flows, customer reviews, help center articles, landing pages, and much more.
Then I asked each LLM to provide structured results; top competitors, top terms used in the industry, top overall term recommendation, and detailed insights in each of these area.

4. Test prompt on different LLM models

My team and I tested the prompt on different deep research and general models in Gemini, Claude, ChatGPT and Wix’s AI custom chat hub.

5. Build UI

Using our partner, Base44, my team and I built a code-less app to display the concept.

6. Launch product

We launched our product by presenting to the 170+ Writers, then to the Head of the Product Guild at Wix.

7. Implement into content design workflows

It now takes me 2 minutes (instead of 2 weeks) to perform competitor research, and it’s way more thorough. I integrate this into my content design process for every new product or feature. Numerous writers across Wix are using the product regularly too.

Achievements

Adopted by 170+ writers and reduced research time by 90%
Saves writers 2 weeks of work on each new product
Kickstarted my confidence to build custom chats for daily life - new products at work, productivity improvements, and personal organization