Moved by the Word: More than a Prayer List

Moved by the Word is a mobile-first app designed to revitalise one’s prayer life. Track prayer requests in a single list, displayed in a feed that always shows the next requests to cover - no more overwhelming long list, and no forgetting anyone either.

Features

The app focuses on facilitating a powerful prayer life:

• Smart feed cycling that surfaces requests based on priority and recency of prayer
• Notes thread per request for adding context and updates
• Memory verses using a purpose-tuned spaced repetition system (SRS)
• Verse picker to quickly add content from KJV Bible - no app switching
• Theme support with light, dark, and system modes

Why I Built It

I wanted a simple tool to help me keep track of prayer requests and ensure I was praying for people consistently. I had ideas for how to do something like this, but work actually began on the app when I really felt the need for this myself. Over time I had plans to expand into scripture memory and social features, but as every mission ought to, this starts with prayer.

Technology

Prayer App is built with a modern offline-first stack:

• Vue 3 with Composition API for reactive state management
• SQL.js (SQLite in WebAssembly) for full SQL capabilities in the browser
• Tailwind CSS v4 with custom theming
• Vite for fast development and optimized builds

Development Journey

Early Conception

The idea for a Moved by the Word app began from discussions and prayer over ministry plans with a few friends. We had big ideas of what we would do, from podcasts to shorts to apps to online bookstores, but weren’t sure what the first step should be. I tried prototyping a devotional app which would use dynamic AI-generated questions, with sophisticated prompting, to guide a reader through structured questions as they study the text; however I found I needed to grow in my ability to ask good questions before I could build something that I would find truly useful.

Other ideas I had included a Bible memory system that functioned more like an infinite scrolling social media feed, using a spaced repetition system-based priority score and somehow incorporating a mechanism for users to journal about their engagement with the text along the way. I also considered expanding this out to a social community where users could share their lessons from devotions with one another. I researched and prayed and planned about what technologies to use for this system, and God gave me direction on certain points. However, nothing really got off the ground.

I still kept thinking about what this ministry could look like and how it could start, but as a team we each became quite busy with other valuable things in life and Moved by the Word laid dormant.

The Need for Prayer

Fast forward to January 2026, and AI coding tools continued to advance in capabilities, meaning the friction of building something new became less and less - and the decision of what to build became more and more important. I didn’t feel the need for a new app to be built for devotions, as I found Logos Bible to cover my needs for Bible study from my phone. However, I found that it was difficult for me to keep track of prayer requests, and existing solutions never felt like a great fit for meaningful prayer. So, I decided to prototype a prayer app based on the paradigm I had originally envisioned for scripture memory.

I couldn’t often justify significant development time at my computer, so I decided to try a mobile-based workflow somewhat inspired by Simon Willison’s writing on how he builds his many small hobby projects. I used Codex web to build and Vercel to serve previews, and a Kanban board in Notion to track next tasks. With even a fairly simple PWA prototype I was able to validate the concept met my personal need, so I invested a bit more effort into refining the UX. I found GPT-5.2 perfectly smart enough, but it lacked the taste to make the UI beautiful, and I had also hit my Codex weekly rate limit in roughly a day, so I needed another agent. I turned to Opus 4.5 in Cursor web, and also discovered that Codex web costs about 5x the usage that the CLI or extension does, while Cursor seemed to have no equivalent markup, so I moved to mostly use Cursor agents for remote coding. I did a pass of UI refinement from my laptop to use some Warp credits before the monthly reset, and moved to Linear for task tracking due to its integrations with cloud agents.

By this point, I had a viable prototype, with much more room for meaningful growth than my other hobby projects thus far. Auth and cloud sync could allow my wife and I to pray over the same requests together, and further list management and sharing features would also be helpful for us and our friends. While the PWA is still shareable, building a proper mobile app and distributing to app stores would make a smoother experience with less distractions and better notification options. Stay tuned for updates!

Expanding Territory

There were so many possibilities of features to add, but I had to be laser-focused on what’s most important in order to make progress. I needed a list view to manage an expanding list of prayers without having to swipe through them individually, as well as a mechanism for viewing archived and answered prayers. This came in the form of a list view, which required rethinking navigation. I was also eager to add scripture memory and pushed to squeeze that into the same update.

I did significant planning upfront, then let agents build things out, then iterated on things that didn’t look quite right or work as intended. It took some iteration to get strong continuity between the prayer and memory feeds, especially considering the progress dots needed to track queues with quite different behaviour. Those dots cost me a lot more time than anticipated - but especially for memory verses, I found it quite valuable for visualising where I was in the queue. I also had to tackle a bit of agentic laziness with vendoring the KJV content in JSON form from an open source repo; even though I had given all the instructions, I had to handhold a bit more to get everything connected with complete integration. In order to address this agent behaviour, I’m working on better fleshed-out agentic browser testing, by identifying user stories and using agent browser to report on behaviour from a user’s perspective rather than a developer’s perspective.

After shipping, I pushed a few minor fixes and algorithm tweaks, and now I’m setting my eyes on user accounts and moving beyond local-only.