How The Way Version Was Built: AI Research and a Public GitHub Record
The Way Version is built in public. That is one of the easiest ways to understand what makes the project different.
The current Bible, earlier versions, research notes, open questions, change logs, and quality checks are kept in a public GitHub repository. Readers do not have to see only the finished sentence. They can look at how a question came up, what was researched, what changed, and what was left alone.
AI is part of that process. It helps the project compare large amounts of text, find patterns, surface possible problems, and test consistency. But AI is not treated as the final authority. Its suggestions are checked against source texts, language resources, scholarship, and the project’s own history.
The basic method
The project asks a simple question over and over: What does the earliest recoverable evidence suggest this passage was communicating?
That can mean looking at Hebrew, Aramaic, Syriac, Greek, historical context, or the way a word is used elsewhere. The answer depends on the passage.
For the Hebrew Bible, Hebrew is primary. For Yeshua’s sayings and Gospel passages shaped by Semitic speech, Aramaic, Syriac, the Peshitta tradition, and the wider Semitic setting are very important. Greek witnesses still matter. In later New Testament books, Greek evidence may carry more weight when that is where the surviving textual case is strongest.
The project started with a public-domain English Bible
The Way did not begin from a blank page. It began from the public-domain World English Bible.
From there, the text was researched and revised across names, wording, repeated terms, gender, idiom, and source-language differences.
That is why some parts of The Way feel familiar while other lines look very different.
What AI actually does here
AI can search the whole Bible much faster than a person can do by hand. It can find every place a name appears, compare how a word is translated in hundreds of verses, and flag places where a rule works in one passage but fails in another.
For example, a person may notice that one verse says “Ruach” and another says “Spirit.” AI can find every related case, group them, and show where a simple global replacement would create mistakes.
The same process can be used for names, divine pronouns, “turn back,” human gender language, repeated titles, and many other questions.
The important part is the next step: the project checks the result instead of assuming the model is right.
How a change moves through the project
Step | What happens |
|---|---|
1. A question appears | A name, phrase, pronoun, or repeated term looks inconsistent or may hide older meaning. |
2. Research begins | AI-assisted searches and source comparison look at the issue across the whole text. |
3. The cases are narrowed | Changes are reviewed verse by verse instead of being applied blindly everywhere. |
4. Exceptions stay visible | Human, ambiguous, or differently sourced cases can be left unchanged. |
5. The result is checked | The EPUB and related files are tested so the approved changes do not break the rest of the Bible. |
A real example: divine pronouns
One public GitHub review looked at masculine pronouns used for God across the whole Bible. The rule was not “remove every he, him, or his.” Human references, Yeshua references, and unclear cases were protected.
The review produced a final ledger of 1,126 verse-level changes across 51 books. The project then checked that all 66 books were still present, chapter structure remained intact, links worked, and the approved verses matched the final EPUB.
The useful part is not the number by itself. The useful part is that the work is documented. A reader can see that the project did not simply run a search-and-replace.
Some questions stay open
Not every issue has one obvious answer. The GitHub repository contains open questions about things such as when to use Ruach or Spirit and how to handle Father and Cosmic Parent language in different New Testament settings.
That is intentional. Sometimes keeping a question open is more accurate than forcing every verse into one rule.
Names are only the beginning
The easiest thing to notice in The Way is names such as Yeshua, Yochanan, Ya’akov, Marym, Dawid, YHWH, Elohim, and Ruach.
But the same method can affect other parts of a verse.
Acts 1:16 can say “brothers and sisters” when a mixed community is in view. Matthew 4:17 reads “Turn back” instead of only using the familiar religious word “repent.” Other passages keep distinctions between words that English Bibles sometimes group under one term.
The point is not to make the Bible feel unfamiliar. The point is to bring more of the older language and setting back into view when the evidence supports it.
Why the public record matters
Most readers never get to see how a Bible translation changed from draft to final text. The Way tries to make that process easier to inspect.
GitHub records questions, research, earlier forms, changes, and QA. AI makes large-scale comparison possible. Source texts and scholarship help keep the research grounded.
That does not mean every decision is beyond debate. It means the debate has somewhere to point.
The short version
The Way Version is a 66-book contemporary English Bible built through an ongoing, documented search for earlier meaning.
AI helps broaden the research. Hebrew, Aramaic, Syriac, Greek, and historical evidence help test it. GitHub keeps the work public and reviewable.
That is the method behind the names, wording, and other differences readers see on the page.
Read the Bible With Original Names
Explore a 66-book Bible shaped by original-language research and a public translation record.
Free browser reader and EPUB. No purchase required.