Questionnaire App
See the content in its natural habitat →
The Goal
We learned from our research that LLMs like content with relevant apps built in. So we decided to test the concept on our pillar guide to Pipeline Health. The goal of this project is to…
- Get citations from LLMs
- Increase active users on the website
- Increase engagement time on the website
The Plan
This test project is purely an engagement and brand play. We made the conscious decision not to gate even the output of the app. This is to test a theory that if we give away valuable content, enough of the users will want to continue to work with us. The reciprocity principle. So, the plan was to use AI to code the app, pair it with pillar content, and embed it with the CMS.
The Process
This has been a vibe coding learning experience–nay, learning adventure for me.
I used Claude to code the app. I first had it generate a Product Requirement Document based on the pillar content itself. Then I reviewed the PRD, suggested a few pages, and Claude executed the plan. Because I gave the model so much context in the documentation, it made a pretty darn good V1. I gave it some feedback, and within a few short iterations, the app was ready to be released into the wild.
Once I had the HTML, I wasn’t 100% sure how to embed it on the site. I thought a simple custom code element would do the trick, but no dice; it was well over Webflow’s 50K character limit. So I did some troubleshooting and research, and I learned that the best way to get the app live would be to host it on GitHub. I already had an account and an empty repo, so I added my code, made a few updates so it would automatically resize for different screens, and voila. It all worked.
Once it was live, I did some testing to make sure nothing was broken. I didn’t find any bugs.
The Results
It’s still very early days. At the time I’m writing this, the app has been live for less than a week. I really hope I remember to come back here and give an update. Fingers are crossed.
Learnings
I had dabbled in so-called “vibe coding” before, but this was my first successful soup-to-nuts implementation on the website. If all goes according to plan, it will be the first of many. But the main thing I learned was the how of getting this done. It was a lot of fun, and I’m looking forward to learning more.
The App
I recommend you go to the website and play around with it yourself. But this screen recording is a decent demo if you don’t have the time:
Sample Pipeline Health Report
This is the PDF output I got from the app when I tested it for the recording above. It’s not perfect, but it’s still pretty cool.
Leave a Reply