RedLetters

The corpus

RedLetters works from one fixed source: the spoken words of Jesus in Matthew, Mark, Luke, and John, King James Version. Nothing else enters the analysis — no epistles, no commentary, no narrator descriptions, no retrospective quotations by others.

The corpus never expands. That constraint is the product. As the obvious observations are exhausted, the findings get harder to find and more interesting to read.

Source: CrossWire SWORD KJV module, Words of Christ markup — tracing to Louis Klopsch's 1901 Red Letter KJV. Red-letter boundaries are editorial judgments, not manuscript features. Disputed passages (e.g. John 3:16–21) follow the Klopsch/CrossWire convention. 1,952 verses. Public domain.

What a finding is

A finding is a specific, sourced, defensible observation about the text. It observes — it does not conclude. It does not tell you what the passage means for your life, what you should do, or what God wants. It tells you something about the words themselves that you may not have noticed.

Every finding is traceable to specific verses. Every quantitative claim has a stored verification run. If a finding turns out to be wrong, a dated correction note appears publicly — the original is never silently overwritten.

How it works

Every finding is generated and reviewed entirely by AI. There is no human approval gate.

Generation. The research engine receives the full corpus context, the depth setting, and the list of already-published and rejected topics. It produces a candidate finding as structured JSON — title, hook, insight, source verses, evidence notes, confidence, and safety level.

Mechanical validation. Before any AI review, three validators run deterministically: every source verse is checked against the red-letter corpus, quantitative claims must have stored verification output, and required fields are enforced. A narrator verse in source_verses fails automatically.

Adversarial review. A second AI call evaluates novelty, clarity, conceptual overreach, and product fit. It approves or rejects with a single reason. Mechanics are handled upstream — the reviewer judges only whether the finding is genuinely interesting and honest.

Publication. Verified or Probable findings at safety levels 1–3 publish the next morning. Interpretive, Speculative, or higher-safety findings go to a holding queue and publish weekly.

Confidence and safety

Verified

Countable and reproducible. A second reader with the same corpus arrives at the same result.

Probable

Strongly supported, with one minor interpretive step.

Interpretive

A plausible reading — not the only defensible one.

Speculative

A hypothesis clearly bounded by the text and labeled as such.

Safety levels run 1–5. Level 1 is a purely textual observation with no doctrinal implication. Level 5 is highly speculative and always labeled. Every published finding displays both.

What this is not

RedLetters findings are not pastoral advice, theological conclusions, or the position of any church or denomination. They are research observations from a fixed text, generated and evaluated by AI, and disclosed with confidence levels and source verses on every finding.

Methodology