Legal Disclaimer

Exit Protocol is NOT a law firm, and our employees are NOT acting as your attorneys. The information provided by this platform, including forensic reports, asset graphs, AI-generated narratives, impeachment analyses, and communication analysis, is for informational and organizational purposes only.

Exit Protocol's platform operates at two distinct processing tiers with fundamentally different evidentiary implications:

• Deterministic Core: The LIBR tracing engine, Impeachment Engine, Entity Resolution Graph, and Sentinel Engine use fixed mathematical rules applied to structured data. These modules do not involve AI or probabilistic models. Their outputs are transparent, testable, and replicable — designed to circumvent the "black box" problem that renders generative AI outputs potentially inadmissible under the Daubert standard and Federal Rules of Evidence.
• AI-Assisted Features: The Grand Strategist, AI Judge, BIFF Filter, Judicial Temperature, Leakage Analysis, and communication tools use Google Gemini 2.0 Flash. These are probabilistic models that may produce inaccurate or contextually inappropriate outputs. AI outputs do not carry the same evidentiary standing as the deterministic core.

The spatial-grid OCR pipeline uses Azure Document Intelligence (cloud) and Surya OCR (local fallback) to extract tabular data from uploaded PDF bank statements. While the system analyzes the geometric structure of pages to reconstruct ledgers from degraded, skewed, or multi-generational scans:

• OCR Errors: The system may misread characters (e.g., reading a "6" as an "8") on poor-quality PDF scans. You are responsible for verifying all extracted financial figures against original bank statements.
• Source Data Integrity: The platform processes data as received. Errors in source documents will propagate through all downstream calculations. Exit Protocol does not verify the authenticity or accuracy of uploaded documents — we verify only that they have not been altered post-upload via SHA-256 hashing.

The Lowest Intermediate Balance Rule (LIBR) engine applies the deterministic formula S_t = min(S_t-1, B_t) to trace commingled separate property claims. This methodology is constructed upon established appellate jurisprudence, including See v. See (1966) 64 Cal.2d 778, In re Marriage of Mix (1975) 14 Cal.3d 604, Pereira v. Pereira (1909), and Van Camp v. Van Camp (1921).

The engine is immune to the "replenishment fallacy" — once a funds exhaustion event permanently reduces the separate property claim, subsequent community deposits cannot restore it. While mathematically precise, the output requires a qualified professional to validate the accuracy of underlying input data.

The Impeachment Engine cross-references thousands of imported transactions against figures stated in a party's sworn financial disclosure or affidavit. If a litigant claims a monthly expense of $5,000 but the engine calculates a moving average of $14,000 in outgoing debits, the system flags the contradiction.

The Entity Resolution Graph maps fund transfers through intermediaries, shell entities, cryptocurrency nodes, and offshore endpoints using link analysis. The AI-powered Leakage Analysis detects gaps between reported income and estimated lifestyle spending.

• A mapped transfer path does not independently verify ownership or legal character of destination accounts.
• A negative income-vs-spending discrepancy does not constitute proof of hidden income — it is a mathematical indicator warranting further forensic investigation.
• Entity graph outputs should be used to guide targeted subpoenas and discovery requests, not as standalone evidence.

The Sentinel Engine uses deterministic, rule-based logic to detect potentially suspicious financial activity — including large transfers (>$500), cryptocurrency exchange purchases, gambling transactions, and account security changes.

The Grand Strategist translates LIBR output into structured Statements of Facts and dissipation arguments using AI. The AI Judge, BIFF Filter, and Judicial Temperature modules process communication text for sentiment analysis, tone rewriting, and contradiction detection.

• Grand Strategist outputs are draft documents — not court-ready submissions. All must be reviewed and approved by a licensed attorney.
• AI-rewritten messages should always be reviewed and edited before sending to opposing counsel or the court.
• Judicial Temperature scores are simulations and do not predict actual court outcomes.

The Shadow Report inspects the underlying layer structure of uploaded documents to detect data leakage in three primary vectors:

• Ghost Text Layers: Hidden searchable text beneath unflatted redaction blocks in civilian-grade PDF editors.
• XMP Metadata Leaks: Embedded author, producer software, and creation timestamps that reveal document origin information.
• EXIF Image Fingerprinting: GPS coordinates, camera models, and exact creation dates extracted from embedded images.

While the LIBR calculator adheres to the mathematical standards established in See v. See (1966) and Cunningham v. Brown (1924), and while the deterministic core is designed to satisfy the Daubert standard and Federal Rules of Evidence, a computer-generated report may not be automatically admissible in your jurisdiction without a human expert witness to authenticate it.

Our chain-of-custody features use dual-hash verification (SHA-256 + MD5) to prove that a file has not been altered since upload. Certified Forensic Reports are sealed with a unique SHA-256 hash — any modification to the report, even a single digit, will invalidate the cryptographic seal. However, this does not prove that the original file was authentic or unaltered prior to upload.

When you connect third-party services such as Clio Manage or Plaid, Exit Protocol retrieves data using read-only OAuth 2.0 access.

• Data Accuracy: Exit Protocol is not responsible for the accuracy, completeness, or timeliness of data provided by Clio, Plaid, or your financial institution. Errors in source data will propagate through our calculations.
• Service Availability: We are not responsible for downtime, API changes, or service disruptions caused by third-party providers.