Every tool in TheoLink, in plain language.
Explore every feature of TheoLink explained in clear, accessible language — no jargon, no technical details. Just what each tool does and why it matters for your research. Looking for technical details?
Every text in TheoLink goes through the same pipeline, from a verified public-domain source file to a deeply-analyzed, cross-linked passage you can search semantically. Here’s every step in plain language. Want the technical version?
Every text starts as a verified, public-domain scholarly translation. We pull from Project Gutenberg, Internet Archive, and Sacred-Texts.com — the same editions cited in academic papers. Each source is logged with its translator, year, original language, and download URL, so you can reproduce or verify any of it. Today: 380+ declared texts across 25+ traditions, with more added every week.
This is where most AI corpora go wrong, and it’s why TheoLink exists. Other systems chop sacred texts into 700-token blobs, destroying the verse, sutta, ayah, or stanza boundaries the author actually wrote. We refuse. Each text type gets a parser that knows its native structure: Bible chapter:verse, Quran surah:ayah, Bhagavad Gita chapter:verse, Pyramid Texts utterance+line, Gospel of Thomas logia, Tao Te Ching’s 81 chapters, Iliad book+line, hadith trees, and so on. The smallest author-defined unit becomes a “Passage” with a globally-unique reference like genesis::1:1 or quran::2:255. That reference is stable across every translation of the same work.
When a passage is too long for a single embedding (rare — most aren’t), it’s split at sentence boundaries with overlap, but the original passage identity is preserved. Search returns results at the passage level, never at chopped-up-blob level.
Every passage is sent to Voyage AI’s voyage-3-large model, which converts the text into a 1024-dimensional semantic fingerprint. These fingerprints let us find passages that mean similar things even when the words differ — paraphrases, translations, theological echoes across traditions. Stored in a dedicated vector database for sub-100ms retrieval.
Every passage is then analyzed by Anthropic’s Claude through thirty-five separate focused lenses, each one tuned to extract a different layer of meaning. Naming, symbolism, narrative structure, doctrinal content, embodied practice, register, intertextuality, and cross-tradition convergence. Simpler lenses run on Haiku (cheaper, faster); judgment-heavy ones run on Sonnet. Both go through Anthropic’s Batch API with prompt caching for substantial cost savings, and every extraction carries a confidence score plus model + prompt-version provenance — so an improved prompt can refresh just that lens without redoing the others.
The lenses by category:
Beyond Jung and Eliade and Thompson, there’s a working ontology of trans-traditional patterns scholars actually use — apophatic mystery, the creative word, cosmic sacrifice, the interior kingdom, ego-transcendence, death-rebirth initiation, axis-mundi, breath-as-spirit, name-as-power, image of the divine in the human, surrender as means, the middle way, the divine that wills its own self-disclosure. The perennial-pattern lens checks every passage against this catalog and asks: is this articulating one of these patterns, and how cleanly?
Every match carries:
This is what lets TheoLink surface where religions point at the same things — with honest scoring — AND surface where they genuinely diverge. The middle path between “all religions are the same” (which betrays the seekers) and “every tradition is a sealed island” (which betrays the data). You walk the gradient yourself; the engine surfaces the candidates with calibration.
All extractions land in Neo4j Aura, a managed graph database. Passages link to themes, themes link to other themes across traditions, entities link to the passages they appear in, places link to coordinates on the map. When you click “show me everywhere Sophia is mentioned,” the graph returns the answer instantly.
When you search, we run two lookups in parallel: a vector-similarity search for semantic matches and a full-text keyword search for exact-phrase matches. The two ranked lists are fused (Reciprocal Rank Fusion — the same algorithm used by major hybrid-retrieval systems) and then re-ranked by Voyage AI’s rerank-2.5 cross-encoder. You get both “passages that mean the same thing” and “passages containing the exact phrase,” ordered by genuine relevance.
Whenever we improve a parser or a prompt, we don’t lose data. Each text re-ingests as a single atomic operation: snapshot the existing state, purge the old passages and embeddings, parse, embed, extract, store, then verify nothing was lost. If anything looks wrong (e.g. extraction edges silently dropped to zero), the re-ingest fails loudly instead of corrupting the corpus.
TheoLink runs on Claude (Anthropic), Voyage AI, and Neo4j Aura. None of this would be possible at this depth or this price without them — sincere thanks to all three teams for the infrastructure that lets a small project punch this far above its weight.
Immerse yourself in sacred texts with tools designed for deep reading and comparative study.
Navigate the web of connections between traditions through interactive visualization.
AI-powered analysis with genuine scholarly rigor — every claim traced to its source.
Deep structural analysis grounded in established academic frameworks from comparative mythology, folklore studies, and the cognitive science of religion.
Detect, visualize, and reason about cycles — historical, astronomical, mythological, cultural, and narrative recurrence patterns.
Track how entities, motifs, and narratives transform across traditions and centuries — trajectories, not just snapshots.
Precession mapping, cognitive science of religion, environmental determinism, and fractal pattern detection.
Beyond structure — the felt dimension of sacred texts and the shape of the stories they tell.
Guided learning paths and progressive tools that meet you where you are.
Spatial analysis of how ideas moved through the ancient world.
The system learns, remembers, and generates — not just a tool, but a research partner that knows you.
Install The Archive on any device — desktop, mobile, or tablet — with native app store distribution and full offline support.
Every primary source text currently loaded in TheoLink, grouped by tradition. Live from the lineage database — 551 texts across 23 traditions. See the full corpus roadmap for what’s staged, queued, and deferred — including untranslated canonical gaps in their original languages and copyright-blocked items.
Traditions and texts currently in research and development. These will be added to TheoLink with the same depth of analysis, cross-referencing, and AI-powered pattern detection as our existing corpus. The full roadmap tracks every queued text, every untranslated canonical gap (with original language), and every copyright-blocked item with its standing translator.
Guru Granth Sahib — the living scripture of 1,430 pages composed by six Sikh Gurus and 15 bhagats across Hindu and Muslim traditions.
In DevelopmentTattvartha Sutra, Kalpa Sutra, and Agamas — foundational texts on non-violence, karma, and the nature of reality from one of the world's oldest religions.
In DevelopmentKojiki and Nihon Shoki — the creation myths of Japan, chronicling the divine origins of the islands and the imperial line from Amaterasu.
In DevelopmentThe Analects, Mencius, and the Great Learning — ethical and political philosophy that shaped East Asian civilization for over two millennia.
In DevelopmentThree plans to match how you use the app:
• Free ($0) — 3 AI messages/day, browse the library, explore the graph, view the timeline. Ad-supported. • Scholar ($14.99/month) — 50 AI messages/day, no ads, export your research, 50 research projects, AI text-to-speech. • Oracle ($79.99/month) — 150 Sonnet + 15 Opus deep-analysis messages/day, pay-as-you-go Opus overage, unlimited research projects, API access, and unlimited TTS.
Kitab-i-Aqdas, Kitab-i-Iqan, and the Hidden Words — texts proclaiming the unity of religion and the oneness of humanity.
Ifa divination corpus (Yoruba), compiled oral traditions, and creation narratives from across the African continent.
In Development