Fellwork — deep Bible study, built for the rest of the world

The thesis

Pastors and students in emerging markets are doing serious exegetical work on $200 phones over metered data. The existing stack — Logos, Accordance, BibleWorks — was designed for North American seminarians on broadband with disposable income.

A faithful, accountable, deeply-linked Bible study experience can run on a feature-class device, in any of 15 languages, without a subscription gating the original Greek and Hebrew. So we’re building it.

Built solo. 3 months. No funding.

The data foundation, already in place.

Curated, computed, and aligned data layered over the entire 66-book canon — on commodity hardware, by one founder.

Original‑language foundation

Hebrew & Greek lemmas: 15,110
Word-level alignments (Heb/Grk → English): 1,148,493
LXX (Septuagint) alignments: 325,482
Linguistic senses: 87,833
Sense definitions: 52,777
BHS critical-apparatus entries: 49,825
Masorah notes: 37,822
NETS Septuagint verses: 23,145

Commentary intelligence — 215 volumes ingested

Cross-references extracted: 444,717
Discourse-structure notes: 112,246
Word studies: 70,072
Documented commentator debates: 18,584
Pastoral applications: 14,294
Depth insights: 5,403
Pericope units: 4,507
Book outlines: 2,452

Theological & structural layer

Theological concepts: 41,152
Concept ↔ original-language alignments: 62,839
Word cards: 13,735
Phrased blocks: 8,317
Genre sections: 1,194
Messianic prophecies cross-referenced: 310
Literary structures: 18

Coverage: all 66 books of the canon. ~3M+ curated / derived data points. The faithful translation engine — the user-facing output, with deterministic phrasing and per-stem glosses — is being built on top of this foundation now.

Open infrastructure shipped along the way

Rust crate

magna

Auto-generates a GraphQL API from any Postgres schema. Field-level extension surface (Rust trait, YAML, HTTP webhook, WASM in v0.5). MIT / Apache, honest stability labels, written succession plan.

Permissively-licensed alternative to Postgraphile and Hasura. Powers Fellwork’s own /graph/graphql surface.

github.com/fellwork/magna →

TypeScript

scribe

@scribe/signals — ~1 KB gz reactive primitives. @scribe/arbor — small renderer. Sync semantics, lazy computed, explicit batch. No proxies, no scheduler queue, no global tick.

Built so the Fellwork reader can run smoothly on cheap Android phones in low-bandwidth markets. The whole reactive layer is smaller than a single React component bundle.

github.com/fellwork/scribe →

Who this serves

Pastors and seminarians in the global South who can’t justify Logos’s $400+ / yr pricing but need real exegetical tools.
Students learning biblical Hebrew and Greek in countries where seminary libraries are thin and digital resources are paywalled.
Bible translators working in under-resourced languages who need a faithful, transparent, every-word-traceable English reference.
Anyone on a $200 phone over metered data who has been told that “real” Bible study tools require a desktop.

Where targeted infrastructure investment unlocks the next leg

In-browser dev environments for international contributors. Translators and reviewers in 80+ countries can iterate on the faithful-translation engine without wrestling with Windows / Linux / macOS local setup. Cloud-hosted dev environments remove the single biggest onboarding blocker.
AI-assisted coding agents for the translation-tooling iteration loop. Fellwork’s deterministic language engine is rule-table driven — exactly the workload modern AI coding agents excel at scaling.
Hosted preview deploys for the public-facing scripture data. Cheap, fast preview URLs for every PR mean external scholars and reviewers can validate output before it ships.

Founder

Shane McGuirt

Navy veteran (Retired). ASU Tech Entrepreneurship & Management alum. Currently studying Biblical and Theological Studies at Regent University. Member of Alive Church, Chesapeake, VA. Building Fellwork full-time.

srmcguirt@gmail.com · github.com/fellwork