Parse in Reducto
Run Reducto exactly where it already sits in your workflow: PDFs, scans, tables, images, markdown, or parse JSON.
Reducto Integration
Compatibility Rail
Already using Reducto? Keep it.
PolDex is not asking Reducto customers to switch document parsing tools. If Reducto already turns your files into parse JSON, markdown, or text, send that output to PolDex and get insurance-native structured results from FastScript.
This is an input compatibility surface, not a disclosure of PolDex internal parsing. PolDex keeps its own document-reading stack private while accepting customer-owned Reducto output as one way to enter the insurance truth layer.
Import to PolDex
Post the Reducto parse output, markdown, text, or adapted async event payload to the PolDex Reducto import endpoint.
FastScript adjudicates
PolDex maps the content into insurance-specific schemas, evidence, conflicts, truth states, abstentions, and export artifacts.
Keep your systems
Send the resulting JSON, CSV, XLSX, webhook, or connector event into your existing broker, carrier, MGA, or agent workflow.
API
One import endpoint after your Reducto parse step.
The import endpoint accepts Reducto-style parse output and requires a PolDex schema target. Hardened schemas run normally. Non-hardened schemas remain gated until their benchmark proof passes.
curl -X POST https://poldex-api.plain-water-cd0f.workers.dev/v1/connectors/reducto/import \
-H "x-api-key: pd_live_YOUR_KEY" \
-H "content-type: application/json" \
-d '{
"schema": "coi",
"client_ref": "red-parse-2026-001",
"parse_result": {
"markdown": "CERTIFICATE OF LIABILITY INSURANCE..."
}
}'Why It Exists
Reducto can stay the parse step. PolDex becomes the insurance control step.
The value is additive: keep your current parser, then add PolDex when the output must become evidence-backed insurance data that agents, APIs, exports, and operational systems can trust.
Already built on Reducto
You do not need to rip out a working Reducto deployment just to get insurance-specific structured output.
Need insurance truth
Reducto output becomes the input. FastScript handles schema contracts, evidence expectations, conflicts, and abstention.
Need artifacts
PolDex returns signed JSON, CSV, and XLSX links from the same result object used by API, processor, MCP, and CLI surfaces.
Need separation
PolDex does not disclose its internal document-reading stack, provider routing, prompts, or parsing strategy on this rail.
Retention
PolDex stores the result, not your raw Reducto payload.
The connector records redacted run metadata, schema, customer reference, result links, and event state. The raw Reducto parse output is treated as processing input for the import call, not as the product's stored source-of-truth artifact.
Add insurance truth after Reducto.
Keep the parsing workflow. Add schema control, evidence, conflicts, abstention, and export artifacts.