Quote-vs-invoice line-item comparison across any invoice history. Every variance classified: price increase without authorization, quantity substitution, SKU substitution, items billed not in quote, tariff passthrough without documentation. Output is not a spreadsheet — it is a claims package. Each finding carries the invoice line item, the approved quote reference, the delta amount, and the classification. Lookback engagements: 1-year, 3-year, 5-year. Structured for use with the supplier directly or with counsel in a commercial dispute.

NumPy-driven scenario modeling across multiple damages theories. Lost profits under different but-for assumptions. Business interruption calculations with commodity price anchoring across the damage period. Contract compliance variances — what the contract specified vs. what the invoices show vs. what spot prices were doing during the same window. Output is a structured dataset: every number traceable to a source record, every assumption documented, every sensitivity range computed and labeled. The expert signs it. Synterminal built it. Opposing counsel can cross-examine the methodology. They cannot dispute the arithmetic.

Benford's Law analysis on any financial transaction dataset — invoices, payroll, expense reports, procurement awards. Sequential invoice anomaly detection: missing numbers, duplicate amounts, suspicious rounding patterns. Bid spacing analysis on public procurement datasets: identical amounts, suspiciously tight clustering, non-competitive pricing patterns. Employment discrimination pattern detection: pay equity analysis across demographic cohorts, promotion rate disparities, termination clustering by protected class. Statistical sampling in labor and employment disputes — used to simplify damage analyses and avoid issues with incomplete data. Every output carries the methodology, the significance threshold, and the underlying data.

Web scraping across public records: Secretary of State incorporation filings, UCC financing statements, public bid award databases, court records, property records. Entity resolution logic identifies when two records refer to the same underlying entity — same principal, same address pattern, same registered agent — even when the names differ. Output is a structured entity graph: every node a legal entity, every edge a documented relationship, every connection traceable to its source record. Applicable to fraudulent conveyance matters, trust and estate disputes where assets are hidden, construction fraud with ghost subcontractors, and procurement collusion investigations. The opposing party cannot claim ignorance of a public record the graph cites directly.

The Synterminal document search system — Flask, SQLite, FTS5, semantic search — ingests any document corpus: discovery productions, deposition transcripts, contract repositories, internal communications, financial records. Every page indexed. Full-text search with relevance ranking across the entire corpus. Semantic search finds conceptually related content even when the exact term doesn't appear. Matter-specific deployment: the corpus stays within the engagement, is not shared across matters, and is not retained after the engagement closes. For a solo or small firm, this replaces a $300–$500/month eDiscovery platform subscription with a custom-built, matter-specific tool that costs a fraction of ongoing SaaS fees.

Custom contract data extraction: pull specific clause types, counterparty names, effective dates, pricing terms, renewal triggers, exclusivity provisions, indemnification caps, and governing law across a batch of contracts — any size. Output is a structured table: one row per contract, one column per extracted field, every cell traceable to the source document and page number. For M&A due diligence at boutique firms that cannot justify a Kira or Luminance enterprise license for a single deal — this is the same extraction capability scoped to the engagement at a fraction of the platform cost. For contract compliance disputes — the extracted terms are the baseline against which invoice variances are measured.

Chronology reconstruction from any mix of structured and unstructured sources: emails, transaction records, invoices, contracts, court filings, communications. Events sorted by occurrence date, cross-referenced by recording date, gap-detected against expected sequence. Every event anchored to its source document and page. The bi-temporal model — separating when an event occurred from when it was known — is exactly what damages calculations and chronologies require. A contract breach that occurred on one date but wasn't discovered until a different date is modeled correctly. The output is a verifiable, timestamped chronology ready for the brief, the expert report, or exhibit preparation.

Matter profitability analysis: time recorded vs. time billed vs. time collected, by matter type, practice area, attorney, and client. Billing realization tracking: where does money leave the funnel between work performed and cash collected. Fee benchmarking: what are comparable firms charging for similar matter types in the same jurisdiction. Built as a structured analysis from whatever billing data the firm exports — not a platform subscription, not another $150/month SaaS line item. A single engagement that produces a structured output the firm's leadership can actually use. This is the lowest-barrier door opener for smaller firms — the deliverable is concrete, the stakes are low, the demonstrated value is immediate.