The Bordereaux Analyst reads cedant bordereaux in whatever format they arrive, classifies every line into your ledger's entry codes, and posts a clean, audit-ready feed. Built for the 80% of cedant data that doesn't arrive as a structured feed.
Every line has to land classified to the right entry code, tied to a source document, and explainable at audit. So the people who should be analyzing cedant performance spend the month moving numbers from one rectangle into another.
formats in one month's bordereaux: a clean Excel workbook from one cedant, a PDF with three embedded tables from another, a CSV whose headers shifted again.
Structured-bordereaux standards are real, and we're glad they're growing. The other 80% of cedant data, PDFs, emails, spreadsheets, will keep arriving the way it always has.
Every cedant sends differently, and every quarter at least one changes how.
The senior operator knows Cedant X's Q3 file needs a brokerage override; when she's out, it doesn't happen.
Intake errors compound into quarterly reconciliation, then the loss ratio, then a board deck.
One reinsurance audit we worked through found a 12% profitability uplift hiding in patterns the manual team had been quietly absorbing for years. The real cost isn't the rekeying time; it's the distortion of the numbers your underwriters and CFO decide on.
A single engine wrapped around a per-customer configuration: your entry-code dictionary and your output template.
PDF, Excel, email body, embedded table, scanned attachment: read, structured, with lineage back to the source line.
Written Premium (100), Commission (200), Brokerage (240), Paid Loss (300): every classification carries its rationale, confidence, and source.
A CSV in the exact upload shape your ledger expects; your platform and workflow don't change.
The first deployment with a specialist P&C reinsurer involved 21 entry codes in the initial dictionary; the second reused roughly 80% of that work. The reusable parts compound; the customer-specific parts stay yours.
Every cedant quirk and recurring override becomes a permanent, versioned dictionary entry that doesn't retire when the operator does.
The same team that handles a $50M book can handle a $400M book when intake doesn't grow with cedant count.
Every classification points back to the exact source line with a timestamp and rationale; the answer for any auditor is one click away.
97%+ classification accuracy on production data, with the remainder routed to a human-in-the-loop queue and folded back into the dictionary.
The platform worked, the accounting worked, the audit worked. The gap was between what the cedant sent and what the platform could ingest, and people were filling it manually every month. So we built for that gap: the Analyst sits between your cedants and your ledger, reads what they send and writes what your ledger takes, and gets sharper with every file your team figures out.
Yes. The Analyst is configured against your specific SICS entry-code dictionary and your specific upload template. The engine is platform-aware; the configuration is per-customer. Typical first-cedant-live timeline is 2–6 weeks from kickoff, depending on how many entry codes and cedant formats are in scope at launch.
ADEPT is the industry's structured-bordereaux standard, and it's growing. We expect it to keep growing. But the realistic adoption curve means ADEPT solves a slice of cedant data: the slice where everyone on both sides of the transaction has agreed to the structured feed. Brisc handles the other slice: PDFs, emails, spreadsheets, the long tail of formats your cedants will still be sending in five years. ADEPT and Brisc are complementary, not competitive.
97%+ classification accuracy on bordereaux data, with the remainder routed to a human-in-the-loop queue where the override is logged and folded back into your dictionary. We don't claim 100%. We claim 97%+ with a closed feedback loop and an audit trail that explains every classification, including the ones the model wasn't sure about.
The dictionary is built to absorb format drift. When a cedant changes how their file is shaped, the Analyst flags the change, the operator confirms or corrects the new mapping, and the new mapping becomes part of the permanent record. The first time a cedant changes formats, your team handles it once. After that, the dictionary handles it.
Same engine, separate dictionaries. The classification logic is shared. The entry-code mappings, the regulatory nuances, and the output templates are configured per legal entity. Dual-deployment reinsurers run one Analyst, configured twice.
Onboarding covers dictionary configuration, output-template setup, and integration testing against your SICS instance. Time to first live cedant is 2–6 weeks. After that, each additional cedant onboards faster because the engine and dictionary are already in place.
Send us three real cedant files: any format, any quarter, any level of mess. We'll show you what the Analyst does with them in a working session.
Book a working session