FAQ
Common questions.
Answers to practical questions before you integrate or contact us for a real reason.
Practical answers by surface.
Product
Does PolDex have a dashboard?
No customer account dashboard, login portal, workspace, or workflow system. Developers use the API. Operators can use the Processor review cockpit to inspect evidence, save notes, mark decisions, and export results without turning PolDex into a workflow suite.
Is PolDex synchronous?
No. PolDex is asynchronous by design. POST /v1/extract returns 202 Accepted with a job_id. Results arrive via webhook or polling. There is no blocking API call that returns content.
How do results come back?
Via webhook, polling, signed result links, or processor downloads. Developers can provide a webhook_url and receive a signed payload when the job completes. Operators can review evidence, save field notes, mark decisions, and download named JSON, CSV, XLSX, or ZIP outputs from the processor.
Can one schema run return unrelated fields?
No. Results are shaped around the requested schema. A Commercial GL extraction should not return personal auto, homeowners, life, or other unrelated line facts in the customer-facing JSON, CSV, XLSX, webhook, or processor output.
What document families does PolDex support?
56 release-ready schema contracts across 7 insurance families are release-ready, callable, and 99%+ benchmarked across all 56 schema contracts across the insurance universe.
Integration
What does an Idempotency-Key do?
Including an Idempotency-Key header makes the submission safe to retry. If the same key is submitted twice, PolDex returns the original 202 response without creating a duplicate job.
Can I submit a document by upload?
Yes. Developers can use document URLs through the API, while the Processor supports PDF/TXT/MD uploads, pasted text, and document URLs with launch-size limits.
How do I rotate a compromised API key?
POST /v1/keys/rotate creates a new key and invalidates the old one. The new key is shown once. Store it securely. No support contact is needed.
Billing
How does pricing work?
Prepaid credits. No seats, no monthly commitment, no platform fees. Short/simple documents start at 1 credit; larger documents and batches are estimated before processing.
What happens when credits run out?
POST /v1/extract returns 402 Payment Required. No job is created, no partial work begins. Top up your balance and resubmit.
Am I charged for failed extractions?
No. Internal failures - extraction errors, infrastructure faults - release the credit hold. You are not charged for PolDex errors.
Do credits expire?
Credits do not expire. Your balance persists until used.
Security & Data
Do you use customer data for model training?
No. Customer content is not used to train or fine-tune any model by default. The inference provider is contractually prohibited from using customer data for model training under their standard terms.
How long is my data retained?
Job metadata and structured output are retained for 90 days by default. Raw document content, worker disk, and processor temporary source text are cleared after successful extraction. Shorter retention is negotiable for enterprise.
Can I delete a job?
Yes. POST /v1/jobs/{id}/delete initiates removal of structured output from the PolDex datastore. Operational metadata (status, timestamps) may persist for audit purposes.
Enterprise
Can enterprise buyers preload by contract?
Yes. Enterprise preload provides a credit allocation under a contract rather than self-serve top-up. Volume discounts negotiable.
Is a DPA available?
Yes. A Data Processing Agreement is available on request for enterprise buyers and for any customer who requires one for compliance.
Do I need a separate enterprise product?
No. Enterprise use does not change the product surface. Same API, same ledger, same no-dashboard doctrine. What changes is the buying path and contractual wrapper only.
Still have questions?
Contact us for enterprise, security, or procurement queries. Docs cover routine integration questions.