Briefing thesis
Tracing separate property through commingled accounts is a sequential calculation problem bound by judicial tracing standards — not a narrative summarization task. Exit Protocol's V1 path ingests selected financial records, normalizes ledger chronology, runs deterministic LIBR tracing, and exports attorney-reviewable workpapers with source provenance and snapshot integrity. The platform organizes review material; counsel and retained experts determine characterization and final use.
High-conflict litigation still relies on human experts to interpret records, select strategy, and present conclusions. The expensive friction is often upstream: transcribing PDF statements, reconciling running balances, and reproducing the same LIBR walk when assumptions are challenged months later.
Jurisprudential context
When separate property commingles with marital or community funds, a presumption often applies that the account is entirely marital unless adequately traced. The burden rests on the party asserting separate character. Common reference points include:
- See v. See, 64 Cal.2d 778 (1966): California tracing-burden and recordkeeping principles after commingling; the opinion does not define the product's LIBR algorithm.
- In re Marriage of Mix (1975): Record-maintenance burden to overcome commingling presumptions.
- Pereira / Van Camp doctrines: Apportionment frameworks for business growth — outside the default V1 single-claim LIBR export.
Exit Protocol does not resolve legal characterization. It makes the selected calculation settings and ledger math visible for reviewer evaluation under applicable rules.
LIBR mechanics
LIBR applies a protective presumption: withdrawals are modeled against community or tainted funds first, preserving separate character as the last dollars remaining in the account under the selected assumptions.
Let St represent the modeled separate-property claim after processing day t, and Bt the account balance on that day. A simplified formulation:
The replenishment rule
When a balance floor event reduces the modeled separate-property claim, subsequent community deposits do not automatically restore that claim under LIBR. Exit Protocol enforces this deterministically so reviewers can see exactly which transaction produced the floor and how later deposits were treated.
Pipeline architecture (high level)
The V1 workflow is a controlled pipeline from user-selected documents to export — not a background sweep of firm or client systems.
Selected ingestion
User-triggered upload of chosen statements with document identity and hash references stored on the matter record.
Normalization
OCR and document intelligence convert PDF layouts into structured rows; reviewers reconcile opening balance, flows, and closing balance.
Deterministic engine
LIBR runs in reproducible application code with ledger agreement checks against summary totals.
Export packet
Workpaper PDF includes disclosed calculation settings, source provenance links, methodology notes, and snapshot SHA-256 at export time.
Under the hood, the application uses a Django monolith with PostgreSQL for ledger storage and asynchronous task processing for heavier document jobs. Teams with local deployment requirements can run containerized instances subject to their own security review.
Snapshot integrity and verification
Reviewers need to confirm that an exported workpaper matches the inputs and calculation state at generation time. Exit Protocol records snapshot hashes at export and supports exact file-hash verification through the public verification surface — conservative matching only, no filename or partial-hash shortcuts.
Integrity controls support auditability; they do not constitute an admissibility guarantee or expert certification. Daubert and similar standards remain matter-specific determinations for counsel and the court.
V1 scope and known boundaries
The default public path is single-claim LIBR workpapers from selected bank and brokerage records. The following require separate review and are not presented as the default V1 export:
- Multi-claim pro-rata tracing across multiple separate-property sources.
- Automated entity-resolution or crypto chain analysis beyond selected record normalization.
- Generative drafting of legal conclusions or expert opinions.
- Silent bulk ingestion without explicit user document selection.