I started with a simple idea: build a bookstore offering a carefully curated collection of classic and well-respected Christian works from across traditions.

I quickly discovered that deciding what to include was far more complicated than I had imagined. I was relatively new to the literature and, at the time, had read works by only about thirty authors. I began thinking about how I might build a useful reference tool for cataloging and sharing what I found.

Once I recognized how theological ideas develop from one generation to the next, a timeline seemed like the natural format.

Building the Timeline

I initially used ChatGPT to create a basic proof of concept. It did not go well, so I set the project aside.

A few months later, while using Claude Code for another project, I decided to let it take a crack at the timeline. Claude’s proof of concept worked. Over the next several months—this remains a part-time endeavor—I continued adding features to what you now see on the website. I also greatly expanded the administrative tools used to build and manage the database of authors, works, and editions.

Altogether, roughly 30,000 lines of code now support the project, and nearly all of that code was written with the assistance of AI.

The data, however, is another story.

Building the Database

Encouraged by the success of the coding work, I asked three different AI engines to identify the top 100 authors of Christian-related works. I merged those lists and added the authors to the initial database.

I then asked AI to identify works associated with those authors. That process revealed a significant gap in my own knowledge of early Christian literature. Much of it does not consist of books in the modern sense, but of letters, sermons, treatises, commentaries, fragments, and other forms of writing.

After loading a number of works, I asked AI to identify specific editions that could be offered freely as public-domain texts, while also preparing links to paid editions of works still under copyright.

That is where the wheels fell off.

I encountered “verified” links that did not work, editions that did not correspond to the works they supposedly represented, works assigned to the wrong authors, and editions attributed to entirely different titles.

So I backed up and started the data process again.

Where the Project Stands Today

I now have several more specific research and validation tools available for building the database.

I have personally reviewed every author currently included—395 at the time of this writing. I also used AI-assisted research to estimate the level of online interest in each author and used those results to organize the authors into ten research tiers.

I, along with hopefully a few others, will now work through the authors one by one, reviewing their works and the available editions.

The database currently contains just under 3,000 works, and I expect that number to continue growing. The number of authors will also grow as I encounter additional names in the books and historical sources I am reading.

Is the information on the timeline now 100 percent accurate?

Sadly, no.

To make the project available while it is still being developed, I have taken a “close enough for now” approach. A more thorough review of approximately 3,000 works—and what may eventually become close to 10,000 editions—will require considerable time.

What I Have Learned About AI

AI is a tool. It serves the researcher; it does not replace one.

It can expand the researcher’s reach, automate repetitive tasks, identify possible connections, and help build tools that would otherwise be beyond the researcher’s technical ability.

I certainly did not have the time to write 30,000 lines of code just to get this project started. Even with its current errors and omissions, I hope the visual presentation of nearly 2,000 years of Christian thought has value.

Your knowledge and input can help improve that value for everyone who uses the site. Corrections, missing authors, overlooked works, and better editions are all welcome.

Where do I stand today?

AI is excellent—with guidance—at building tools and automating portions of the research process. It is far less dependable at producing clean, verified data.

I will continue testing and evaluating new tools. For now, however, AI will be used where it is strongest—and only where it is strongest.

 

Oh, and one other note: the number of authors, works and editions I mention in this article are higher than what is available online most of the time. When I find data that is incomplete or questionable, I hide it until verified.