Version 4 — Tonight’s Build

Doctrine Engine — AI-Powered Intelligence Engines
A Family of AI Intelligence Engines — Patent Pending

One Architecture.
Infinite Intelligence.

It reads everything. Misses nothing.

Doctrine Engine is a platform of purpose-built AI intelligence engines. Finance, Thesis, Novel, Legal, and beyond — each one delivers forensic-grade analysis at a fraction of human cost and time.

$50,000–$200,000 $17 per full analysis run
Finance Engine · Acme Service LLC · Analysis Running
Full Ingestion
144 documents · 0 truncated · 200K tokens
Done
Corpus Build
23,008 facts extracted · 145 citations
Done
Cross-Reference
Scanning contradictions across all sources…
Running
Dual-Gate Verify + Report
13-chapter forensic analysis · PDF + DOCX
Pending
Contradiction: Purchase option $6M vs $2.975M
Identity confirmed: Jane Doe — 0.94 confidence
Undisclosed debt: $80K personal loans not in PFS
Confidence: 0.94 · 145 citations · $17.43 · 50 min end-to-end
✓ $17/run 50 min Patent Pending
Patent Pending — Corpus Memory Architecture
11 Patent Claims Identified
$17 vs $50,000–$200,000 Human Equivalent
144 Documents · 23,008 Facts · One Run
$17Cost per analysis
$50K–200KHuman equivalent cost
0Truncated documents
200KToken context window

The Business Case

Same output. 3,000× cheaper.

Forensic financial due diligence — human analysts vs. Doctrine Engine.

Traditional Due Diligence
$50K–$200K
3–6 weeks · team of 4–8 analysts
Per engagement$50,000–$200,000
Time to report3–6 weeks
Docs reviewed by partners~20%
Citations per reportVaries
VS
Doctrine Finance Engine
$17/run
50 minutes · fully automated
Per engagement$17.43
Time to report50 minutes
Docs reviewed100% — all pages
Citations per report145 verified
144+
Documents Processed
simultaneously, zero truncation
23K
Facts Extracted
per analysis run
145
Source Citations
verified, not hallucinated
47
V4 Build Items
38 complete · finishing tonight

The Ecosystem

13 engines. One architecture.

Doctrine Engine is a family of AI intelligence tools, all powered by the same Corpus Memory Architecture. Each engine shares the same foundation — ingestion, cross-reference, verification — applied to a different domain.

Doctrine
Engine
Finance
● Live
Finance Engine
Thesis
● Live
Thesis Engine
Novel
◎ Beta
Novel Engine
Legal
○ Coming Soon
Legal Engine
Coding
○ Coming Soon
Coding Engine
Learning
○ Coming Soon
Learning Engine
Medical
Real Estate
Contract
Compliance
Intelligence
Strategy
Live
Coming Soon
Future
💰
Finance
Live
📚
Thesis
Live
✍️
Novel
Beta
⚖️
Legal
Soon
💻
Coding
Soon
🎓
Learning
Soon
🏥
Medical
Future
🏠
Real Estate
Future
🧠
Strategy
Future

The Platform

Every domain. One architecture.

The Corpus Memory Architecture at the core of Doctrine Engine is domain-agnostic. The same ingestion, cross-reference, and verification framework that processes 100 financial documents can process 100 legal briefs, research papers, or novel chapters.

Patent Pending
Live
Finance Engine
Forensic due diligence for commercial lenders, bankruptcy attorneys, and turnaround advisors
Patent Pending
Live
Thesis Engine
Investment thesis generation from research corpora — with citations, verification, and multi-format export
Patent Pending
Beta
Novel Engine
Long-form narrative intelligence — chapters, arcs, characters, and full manuscripts from structured prompts
Coming Soon
Legal Engine
Case strategy, brief analysis, statute cross-reference, and discovery document review at machine speed
Coming Soon
Coding Engine
Architecture review, security audit, codebase analysis, and technical due diligence from repository to report
Coming Soon
Education Engine
Curriculum generation, learning gap analysis, and adaptive content creation from academic corpora
Coming Soon
Medical Engine
Clinical document analysis, treatment protocol cross-reference, and research corpus synthesis
Coming Soon
Real Estate Engine
Property due diligence — title, lien, zoning, lease, and environmental analysis from the full document package
Coming Soon
Contract Engine
Contract review, obligation mapping, risk clause extraction, and cross-party commitment analysis
Coming Soon
Compliance Engine
Regulatory framework mapping, audit trail generation, and compliance gap analysis across jurisdictions
Coming Soon
Intelligence Engine
OSINT aggregation, competitive landscape analysis, and open-source intelligence synthesis
Coming Soon
Strategy Engine
Business strategy synthesis, market analysis, and decision-matrix generation from organizational corpora
Live
Finance Engine

Forensic due diligence at machine depth. The Finance Engine reads an entire deal package — tax returns, bank statements, UCC filings, leases, personal guarantees, court records — and delivers a structured analysis with 145+ citations, contradiction flags, and confidence scores on every finding.

200,000 token context — entire deal package simultaneously, zero truncation
12 financial ratios computed deterministically in <1 second — no LLM math
§547 preference window calculator + §548 fraudulent transfer analysis (Patent Pending)
Benford’s Law statistical fraud detection + dual fact-check gates (deterministic + Opus reasoning)
Confidence scoring 🟢≥0.85 / 🟡0.65–0.84 / 🔴<0.65 propagated from source through every stage
3 deliverables: 13-chapter Financial Analysis + Strategy Memo + 6-chart package (PDF + DOCX)
4-stage decision tree: viability → debt structure → kill signals → recommendation
✓ $17/run 50 min end-to-end 12 ratios · 4-stage decision tree V4: 38/47 items complete 3 Patent Pending claims — §547/§548/Corpus
Finance Engine · Deal Analysis · Acme Service LLC
§1 IDENTITY ✓ Jane Doe — confidence: 0.94 [STRONG] [settlement_doc.pdf p.1] [tax_return.pdf p.1] §3 BALANCE SHEET ✓ First National Bank lien $3,440,000 — 4 sources agree ✗ CONTRADICTION: Purchase option $6M [APA p.1] vs $2.975M [Settlement p.1] §5 UNDISCLOSED DEBT ✗ Capital Funding Co. $300K → $389,999 not in PFS ✗ Personal Lender A personal $50K — undisclosed ✗ Personal Lender B personal $30K — undisclosed §7 PUBLIC RECORDS ✓ Property 1234 Oak Street confirmed ⚠ Related entities: Acme Holding, Acme Motors, Acme Wholesale Confidence: 0.94 · 145 citations · $17.43 total cost
Live
Thesis Engine

From raw research documents to a fully-cited investment thesis — with verification, rewrite, and multi-format export. The Thesis Engine processes research corpora of any size, extracts and cross-references claims, verifies citations against live sources, and produces a publication-ready document.

6-stage pipeline: ingest → extract → Corpus → generate → rewrite → assemble
Citation verification against live web sources — not just document text
Side-by-side diff view for rewrite tracking
Multi-format export: DOCX, PDF, Markdown, HTML
Information gaps auto-identified and flagged per chapter
Revision tracking with max iteration controls
✓ 6-stage pipeline Citation verification layer Patent Pending — Multi-Pass Thesis Architecture
Thesis Engine · Cannabis Market Analysis
STAGE 3 — BIBLE COMPLETE ✓ 47 documents ingested · 312 citations extracted ✓ 94 claims cross-referenced · 8 contradictions found STAGE 5 — REWRITE Chapter 4: Market Sizing Revision 2 of 5 · coherence: 0.91 ⚠ Gap: 2028 projections — source needed STAGE 6.5 — CITATION VERIFY ✓ 89 claims verified · 3 discrepancies ✗ Claim: “$47B market by 2026” — source contradicts [Source now shows $38B — flagged for review] OUTPUT READY ✓ DOCX · PDF · Markdown · HTML export ready
Beta
Novel Engine

Long-form narrative intelligence. The Novel Engine builds complete manuscripts — chapter by chapter, arc by arc — from structured creative prompts, maintaining character consistency, thematic coherence, and narrative voice across 100,000+ words. The same Corpus Memory Architecture that holds a deal package holds an entire novel.

Character tracking and consistency enforcement across all chapters
Thematic arc maintenance — setup/payoff mapped across full manuscript
Style and voice coherence via Corpus-anchored rewrite
Chapter-by-chapter generation with continuity validation
TTS audiobook export — narrated manuscript delivery
Multi-format output: DOCX, PDF, ebook-ready
✓ 100K+ word manuscripts TTS audiobook export Patent Pending — Narrative Corpus Architecture
Novel Engine · Mythos · Chapter 18 of 24
CHARACTER CONTINUITY CHECK ✓ Marcus Hale — arc consistent ch.1→18 ✓ Elena Voss — motivation unchanged from ch.6 ⚠ Dr. Reyes — eye color contradiction (ch.3: gray, ch.17: brown) NARRATIVE ARC Setup (ch.1-8) ████████ RESOLVED Rising (ch.9-15) ████████ COMPLETE Climax (ch.16-20) ████░░░░ IN PROGRESS CHAPTER 18 COMPLETE ✓ 4,847 words · Voice coherence: 0.93 ✓ Thematic echo: “recursion” — 3 instances planted Total manuscript: 87,230 words · 18 of 24 chapters

The Architecture

One Architecture. Infinite Engines.

Every Doctrine Engine product inherits the same patented backbone — document ingestion, Corpus Memory Architecture, dual fact-check gates, confidence scoring, export pipeline. New engines are hours to build, not months. The backbone is the moat.

Doctrine Core
Shared Backbone
Document ingestion pipeline (watcher → parse → extract)
📖Corpus Memory Architecture — the patent (recursive multi-pass compression)
🔗Chapter generation framework (concept → outline → corpus → rewrite → assemble)
🔒Dual fact-check gates (Gate 1: deterministic · Gate 2: Opus reasoning)
📊Confidence scoring — 🟢≥0.85 / 🟡0.65–0.84 / 🔴<0.65 — propagated end-to-end
📤Export pipeline — PDF, DOCX, HTML, TTS audiobook
🔄Self-improvement loop — every run refines future prompts
📡Pipeline telemetry — cost gates, stage guards, state machine
The Moat
“New engine development time: hours, not months. Every vertical engine inherits the full backbone — ingestion, memory, verification, export — and adds only domain-specific prompts, math, and APIs. The backbone is the product. Engines are the applications.”
Finance Engine
12 ratios · §547/§548 · Benford’s · cramdown · DIP
Live
Thesis Engine
OSINT enrichment · citation verify · 24 chapters
Live
Novel Engine
Character bible · continuity enforcement · 100K words
Beta
Legal Engine
PACER · CourtListener · citation verify · case strategy
Coming Soon~2hr build
Coding Engine
GitHub · dependency graphs · security audit · tech DD
Coming Soon~2hr build
+
7 More Engines
Medical · Real Estate · Contract · Compliance · Intelligence · Learning · Strategy
Roadmap

The Architecture

The Corpus Memory Architecture — the engine inside every engine.

Every Doctrine Engine product is built on one patented core: the Corpus Memory Architecture. It is the reason Doctrine Engine can read everything, miss nothing, and produce forensic-grade output at a fraction of human cost and time.

📥
01
Full Ingestion
Every page, every line. 200K token context window. Zero truncation. Zero summarization.
📖
02
Corpus Build
Recursive multi-pass extraction. Every fact sourced to its document, page, and line.
🔗
03
Cross-Reference
Every claim checked against all other claims. Contradictions flagged. Gaps identified.
04
Generate
Full Corpus in context. Output anchored to source at every sentence. No hallucination path.
🔒
05
Dual Gates
Two independent verification passes. Confidence scores on every finding. Full audit trail.
Corpus-at-Ingestion Compression Patent Pending
“A method for compressing heterogeneous source document corpora into a structured, cross-referenced fact base prior to generative AI processing, using a recursive multi-pass extraction architecture where intermediate compression artifacts are themselves subject to further compression when aggregate size exceeds model context limits.”
11 Patent Claims Identified
P-01 Corpus-at-Ingestion Compression Architecture
P-02 Dual-Gate Adversarial Fact Verification
P-03 Cross-Document Confidence Scoring System
P-04 Avoidance Action Calculator (§547/§548)
P-05 Auto-Generated Due Diligence Request Letters
+6 Additional claims withheld — under review by IP counsel
Verified Performance
Documents processed (Sample Deal PoC)144
Dollar figures extracted839
Source citations145
Total run time~50 min
Total cost$17.43

The Origin

It started with a question: what if AI could read the whole thing?

Not summarize it. Not pull key excerpts. Read every page of every document, hold them all in context simultaneously, and find what a human analyst — under deadline pressure — would never catch. The answer became Doctrine Engine.

The first proof-of-concept ran on the Sample Deal PoC deal — 144 documents, 100+ pages of financial records, operating agreements, bank statements, court filings, and personal guarantees. In 50 minutes, the Finance Engine produced a 17,000-word forensic analysis with 839 verified dollar figures and 145 source citations.

A restructuring advisor would have charged $17,000 and taken three weeks. Doctrine Engine charged $17.43.

The analysis found three undisclosed personal loans, a $3M+ contradiction between the stated purchase price and the settlement statement, and a three-way ownership structure that the primary borrower had represented as sole ownership in the loan documents.

The origin document — the research and theoretical framework that predicted the Corpus Memory Architecture would work before it was built — is available to verified partners and investors.
📖
The Self-Consuming System — R. Muranaka (2026)
The original manifesto · Available to verified investors · Access below
🎙️
The Original Prompt — Doctrine Engine Genesis
The exact prompt that generated the first pipeline · Patent-sensitive · Investor access only
Why No Existing Product Does This
RAG (Retrieval-Augmented Generation)
Finds relevant chunks at query time. Misses everything it wasn’t asked to look for. Can’t cross-reference without knowing what to cross-reference. Blind to unknown unknowns.
Document Summarization Tools
Summarizes each document separately. No cross-document awareness. No contradiction detection. No provenance chain. Summaries compress the signal out of the data.
Legal AI Assistants (Harvey, etc.)
Single-document focus. Answering questions, not finding facts you didn’t know to ask about. No forensic cross-reference. Built for legal research, not financial analysis.
Doctrine Engine
Ingests the entire corpus before generating a single sentence. Cross-references everything against everything. Contradictions surface automatically. Unknown unknowns become known — with citations.

Investor Access

Full materials for verified partners.

The complete Doctrine Engine investor package — pitch deck, financial projections, competitive analysis, marketing plan, and the original manifesto — is available behind a password issued to verified partners. Contact us to request access.

🔒
Investor Package
Enter your access code to view the complete materials, including the original manifesto, the first proof-of-concept run, and the full patent filing strategy.
🔒
Pitch Deck
Investor Presentation
32 slides · PDF
🔒
Financials
5-Year Projections
18 pages · PDF
🔒
Analysis
Executive Summary
12 pages · PDF
🔒
Market
Competitive Analysis
24 pages · PDF
🔒
Strategy
Marketing Plan
16 pages · PDF
🔒
Playbook
Guerrilla Strategy
28 pages · PDF
DocumentFinancial Projections
DocumentCompetitive Analysis
DocumentMarketing Plan
DocumentThe Manifesto
DocumentGenesis Prompt
Incorrect access code. Contact info@doctrineengine.com to request access.

Ready to run your first deal?

Doctrine Engine is in early access. We’re onboarding a limited number of commercial lenders, attorneys, and turnaround firms now.

Doctrine Engine outputs are for informational purposes only and do not constitute legal, financial, or investment advice. All findings should be reviewed by qualified professionals. Patent applications pending. Unauthorized reproduction of the Corpus Memory Architecture is prohibited.