I kept running into the same problem. I’d be journaling about something I was struggling with, and I’d think “there’s definitely a Bible verse about this.” But I couldn’t remember it. I’d stop writing, open a concordance, search through keywords. By the time I found something, I’d lost my train of thought.

So I thought: what if the verses could just appear as I write?

Building a Solution

I built Contextful Bible. It’s a free web tool that watches what you type and suggests relevant Bible verses in real time. The key was using AI that understands meaning, not just keywords. I wanted it to catch the essence of what I’m saying, not just match individual words.

I also made sure that when you click a verse, you see the whole chapter context. I’ve seen too many verses misused because they were pulled out of context. Every verse lives in a chapter, and you need to see what comes before and after.

My Motivation

I generally like it when I am somehow ‘forced’ to be closer with the word of God. I’ve liked taking theology classes in college and being ‘required’ to read long Old Testament passages. I’m currently taking a discipleship course in my church and similarly enjoy being told to copy over various New Testament passages in a workbook. This project similarly involved interacting with the bible verses during data preprocessing and testing.

This was my first time using marimo. I had enjoyed using Jupyter notebooks a few years ago (2020-2021) while learning Deep Learning for the first time, and kind of missed that particular programming/learning experience. Marimo provides much cleaner and deployment-ready tools for user-facing apps. This was a big factor in my motivation for starting this project.

Lastly, I enjoyed working closely with the state-of-the-art in language model related technologies: vector embeddings(embeddinggemma-300m), vector databases(sqlite-vec), and high-speed inference (Qwen3 model on Cerebras)

Limitations

Because I am fluent in just English and Korean, I have no idea whether the other supported languages are providing equally high-quality outputs. If you are multi-lingual and have feedback, please let me know.

Try It Out

You can use it at contextful-bible.tensorturtle.com. It is free. The inference cost is very low for now so I am offering it as a public service.