FAQ
General
What types of insurance does Silverarrow support?
Silverarrow is designed for B2B commercial insurance underwriting. The platform is especially optimised for casualty and liability lines.
What file formats are supported?
Silverarrow processes PDF documents. Both text-based and scanned PDFs are supported — the platform uses OCR and AI vision fallback for documents without an extractable text layer.
How long does processing take?
Most submissions are fully processed in less than a minute. Processing time depends on the number and size of documents, the complexity of the submission, and whether OCR or vision fallback is needed.
Can I process submissions in languages other than English?
The AI models used for extraction and analysis support multiple languages. The platform interface and risk insight outputs are in English.
Documents
What happens if a required document is missing?
The case is held at Needs Documents status and an automatic document request is sent to the broker. Should the automatic request not yield the necessary documents, the underwriter can upload them manually from the submission detail view.
Can I upload additional documents after processing?
Yes. Documents can be uploaded at any time from the Documents tab. Uploaded documents go through the same classification and extraction pipeline as email attachments.
What if a document is classified incorrectly?
You can manually reclassify any document from the Documents tab. After reclassification, dependent analysis will re-run using the corrected document type.
Analysis
How accurate is the AI extraction?
Extraction accuracy is above 99%. It can slightly vary by document quality and field type. Silverarrow uses confidence indicators to flag uncertain extractions. High-confidence fields typically have zero error rates. Medium and low confidence fields are flagged and should be verified by the underwriter.
What does “not found” mean on a field?
It means the AI could not locate the information in any of the submission documents, suggesting it does not exist. Rather than guessing, Silverarrow reports the field as not found. This is by design — the platform never fabricates data.
Can I edit AI-extracted fields?
Yes. Every field in every risk insight category can be edited. Edits are tracked with full attribution (who changed what, when) and the original AI value is preserved for reference.
Entity Resolution
What is entity resolution?
Entity resolution is the process of verifying the named insured against external registries such as ZEFIX (Swiss commercial register) and other business intelligence sources, as well as tailored customer integrations.
What if the company cannot be found?
If no registry match is found, the case is flagged for manual resolution. You can enter the company details manually to proceed with the analysis.
Architecture & Data
Does Silverarrow duplicate data from our existing systems?
No. Silverarrow is of orchestrating nature. It synchronises with your existing internal systems — CRM, policy database, and pricing software — without creating data silos or duplicating records. Data lives where it belongs.
Where is Silverarrow deployed?
On-premise at the customer for full control. All data and processing remain within your own infrastructure, under your own security measures. Nowhere else does an organisation in the insurance industry face fewer compliance and security obstacles. Each organisation’s environment is fully isolated.
How is multi-tenant isolation handled?
Since each organisation’s environment is fully on-premise, isolation is absolute by design. Users from one organisation cannot access another organisation’s data.
Export
What format is the exported risk analysis?
The risk analysis can be exported as a PDF document containing all risk insight findings, evidence references, edit history, and approval records. At the customer’s request, the risk analysis can also be exported to Guidewire.
Can I re-export a risk analysis?
Yes. You can re-export the risk analysis at any time after approval. If edits were made after the initial export, the new PDF will reflect the latest state.