Industries
AI Document Processing for Reinsurance
Reinsurance faces a unique challenge in document processing: bordereaux arrive as Excel spreadsheets in hundreds of different layouts. Treaty documents combine legal contract language with complex numerical parameters. Loss notifications arrive unstructured and in varying formats. At the same time, regulatory pressure from FINMA, Solvency II, and MaRisk is increasing — and every extraction must be traceable and auditable. feld.ai was built to master exactly this complexity: a sovereign AI platform that understands heterogeneous reinsurance documents, structures them, and feeds them into core systems.
Typical Documents
Bordereaux
Automatically parse Excel spreadsheets with premium and claims data — regardless of layout and column order. Whether premium bordereaux, loss bordereaux, or cash-loss bordereaux: feld.ai recognizes the structure, normalizes column names, and converts the data into a unified target format.
Treaty Documents
Capture contract parameters, coverage limits, reinsurance shares, and clauses in a structured way. feld.ai extracts retention, limit, quota share, event limits, reinstatement conditions, and cancellation terms from both proportional and non-proportional treaties.
Statement of Account
Automatically reconcile statements against treaty data. The platform identifies premium calculations, commissions, loss reserves, and balances — and flags deviations from contractually agreed terms.
Loss Notifications
Extract claims data from unstructured initial notifications and feed them into core systems. feld.ai reads loss descriptions, claim amounts, affected policies, and involved parties from free-text reports, emails, and PDF documents.
What feld.ai Delivers
The central challenge in reinsurance is what the industry calls the "Excel problem": every cedant delivers bordereaux in a different format. Different column names, different order, different units, different date and number formats. Some bordereaux have multi-row headers, others contain subtotals, and others have nested structures across multiple worksheets. Traditional rule-based and template-driven approaches fail in the face of this diversity — every new cedant format requires manual configuration. feld.ai solves this problem through AI-based table recognition: the platform automatically identifies which columns carry which meaning, normalizes the data, and maps it to a unified schema — without manual mapping per cedant.
For treaty documents, feld.ai combines natural language understanding with structured data extraction. The platform reads contract text, identifies relevant clauses, and extracts numerical parameters such as retentions, limits, and quotas. The result is a structured dataset that can be directly imported into treaty administration systems — including a reference to the exact location in the original document. During treaty renewals, feld.ai detects changes compared to the prior-year version and highlights them for review.
Automated reconciliation of Statements of Account against treaty data significantly reduces manual effort in accounting. feld.ai compares premium calculations, commissions, and loss payments against contractually agreed terms and flags discrepancies. For loss notifications, the platform extracts all relevant information from unstructured sources and produces a structured claims report that can be fed directly into the claims management system.
Why feld.ai
Excel Expertise
Automatically recognize and normalize hundreds of different bordereaux layouts. No templates, no manual column mapping. The AI understands what each column means — regardless of label, language, or position. New cedant formats are processed without any configuration effort.
Regulatory Compliance
FINMA, Solvency II, MaRisk — regulatory requirements for reinsurance are strict. feld.ai runs on its own GPU servers in the EU, with no US cloud dependency. No CLOUD Act, no FISA. Data stays under European jurisdiction. On-premises installation available for maximum control.
Audit Trail
Every extraction is fully traceable. Visual grounding shows where each extracted value originates — with a reference to the exact location in the original document. Complete traceability is guaranteed for internal audit, external reviewers, and regulatory requirements.
Related Solutions
Data Extraction from Records — Structurally unlock, index, and search large document archives.
Excel Table Processing — AI recognizes tables in Excel files, understands structures, and normalizes data.
EU-Sovereign Document AI — Document processing without US cloud — on own infrastructure in Austria.
Frequently Asked Questions
What bordereaux formats are supported?
feld.ai processes bordereaux in all common Excel formats — regardless of column order, labels, or structure. Premium bordereaux, loss bordereaux, and cash-loss bordereaux are automatically recognized and normalized into a unified target format. New cedant format? No manual mapping required.
How are treaty parameters extracted?
feld.ai combines natural language understanding with structured data extraction. The platform reads contract text, identifies relevant clauses, and extracts retention, limit, quota share, event limits, reinstatement conditions, and cancellation terms — from both proportional and non-proportional treaties.
Is feld.ai FINMA and Solvency II compliant?
Yes. feld.ai runs on its own GPU servers in the EU, with no US cloud dependency. No CLOUD Act, no FISA. Complete data processing within the EU meets the requirements of FINMA, Solvency II, and MaRisk. On-premises installation is available.
Is there a complete audit trail?
Yes. Every extraction is fully traceable via visual grounding — with a reference to the exact location in the original document. Complete traceability is guaranteed for internal audit, external reviewers, and regulatory requirements.
Can feld.ai be installed on-premises?
Yes. In addition to the EU cloud solution, feld.ai offers a full on-premises installation on your own infrastructure. All models, the processing pipeline, and the data remain fully under your control.