Technical Features

Editing Fields

Every extracted field in the risk insights can be edited directly by the underwriter. When a field is edited:

  • The original AI-extracted value is preserved
  • The edited value is recorded alongside it
  • A timestamp and attribution identify who made the change and when

This creates a complete audit trail of human overrides on AI-analysed content, ensuring full transparency and regulatory compliance.

Confidence Indicators

Each extracted field carries a confidence indicator reflecting the certainty of the AI extraction:

  • High confidence — The extraction was clear and unambiguous
  • Medium confidence — The field was extracted but may benefit from verification
  • Low confidence — The extraction was uncertain; manual review is recommended

Confidence indicators allow underwriters to prioritise their review efficiently, focusing attention where it is most needed.

Entity Resolution

Entity resolution is the process of identifying and verifying the named insured — the company being covered by the insurance policy. When a submission is processed, Silverarrow:

  • Extracts the company name from the submission documents (policy, broker slip, or email body)
  • Searches external registries to find matching entities
  • Scores potential matches based on name similarity and other factors
  • Resolves automatically if a high-confidence match is found, or flags for manual review

When the system encounters multiple matches, no matches, or a low-confidence result, the case is set to Needs Entity Resolution. The underwriter is presented with candidate matches and can select, edit, or enter the entity details manually. Once resolved, processing resumes automatically.

Entity resolution directly affects the quality of several risk insight categories — in particular the named insured profile, sanctions screening, and ESG assessment, all of which are conducted against the resolved entity.

External Registries

Silverarrow queries external registries and data sources via API to verify and enrich named insured data. Examples include:

  • ZEFIX (Swiss Commercial Register) — Verifies legal company name and registration number, registered address, legal form (AG, GmbH, etc.), and active/inactive status for Swiss-domiciled companies
  • Business intelligence registries — Retrieve industry classification, company size and revenue estimates, and corporate hierarchy including parent and subsidiary relationships

The matched entity data is used to populate the relevant risk insight sections without duplicating that data within Silverarrow itself.

Audit Trail

Silverarrow maintains a complete audit trail for every case:

  • When the submission was received
  • Which documents were processed and how they were classified
  • All AI-powered analyses and their evidence references
  • Every manual edit with before/after values and attribution
  • Approval and export timestamps

This audit trail ensures full transparency and regulatory compliance throughout the underwriting process.