A 14-day evaluation — inside your VPC, on your real files
Bring recent applications you've already approved or rejected. Run them through the engine, see the Risk Ready Packet your desk would have received — then compare it to the call you actually made.
No synthetic data, no rebuilt workflow. The sandbox runs on files you've already decided — so the comparison is against ground truth you already trust.
Pick a batch of recent applications you've approved or rejected — a deliberate mix of both. GST returns, bank statements, invoices, KYC — whatever you have, ingested as-is.
Each file goes through a single authenticated call. The engine verifies entity, transaction and financial — two layers, one pass — and hands back a Risk Ready Packet in ~120 seconds.
Put each packet next to the decision your desk actually made. See what we'd have flagged, cleared, and where we'd have saved hours — measured against outcomes you already know.
This is exactly what runs inside the sandbox. Your LOS keeps running the loan — TruStackAPI runs the verification: one call, two layers, a Risk Ready Packet handed back before the file would reach a credit desk.
Scroll to follow a file through the firewall
The sandbox is the production API with a scoped key. The integration you build against it is the same one you ship — nothing to redo when you go live.
We provision a scoped sk_sandbox_… key and a VPC endpoint, with the full API reference.
/v1/verifySend an application plus its supporting documents — one authenticated request hands the whole file to the engine.
Block for the response or take a webhook — score, severity-triaged flags, and the three-part packet come back ready for the desk.
# sandbox · one file · application + documents curl -X POST https://api.trustackapi.com/v1/verify \ -H "Authorization: Bearer $SANDBOX_KEY" \ -H "Content-Type: application/json" \ -d @application.json
{
"id": "app_4471",
"mode": "sandbox",
"score": "B+",
"decision": "review",
"flags": [
{ "severity": "P1", "label": "Hidden EMI detected" }
],
"packet": { "verification_memo", "cam_bridge", "audit_log" }
}
Sandbox keys, full API reference, payload schemas and the severity taxonomy — provisioned the day your evaluation starts.
Not a score in isolation — the full evidence-graded read your desk would open, in three parts.
The risk read in one page — applicant intel, insights, signals, distilled flags.
80–90% of credit-appraisal prep already structured and ready to carry into the CAM.
Every check audit-trailed and regulator-exportable. RBI FREE-AI ready by design.
Everything you need to judge the engine against your own book, with nothing leaving your cloud and no commitment on the other side.
The engine runs inside your cloud. Borrower data, documents and outputs never leave your perimeter.
Scoped API keys, full reference, payload schemas and the severity taxonomy — provisioned day one.
The product cartridge reads your loan type — checks, scoring slabs and STP thresholds tuned to what you actually write.
Run your decided files, line the packets up against your calls, and judge accuracy on outcomes you already know.
A fixed window with no contract attached. You shape the tunable surface; we refine the engine against your feedback.
Every check audit-trailed and regulator-exportable. RBI FREE-AI aligned, DPDP-ready — from the first sandbox call.
Rules decide, AI advises, humans judge — in the sandbox exactly as in production. Nothing you see in the evaluation changes when you go live.
The fastest way to judge a verification engine is to point it at files you've already called — and see where it agrees, where it doesn't, and what it catches early.
Tell us a little about your book and we'll provision a sandbox inside your VPC, with keys and docs, typically within a few working days.
Built and operated by OneTruth Technologies Private Limited.