GingerControl

Customers get two paths. One is an autonomous trade compliance agent. The other is an AI native GTM platform that integrates directly with the customer's ERP and existing systems. There is no connector catalog, no custom tool route and no outbound authenticated action into another system. ERP and TMS integrations come from the AI Agentic System Build, a per client engagement, not from the platform. The API can be wired into SAP GTS, Oracle GTM or NetSuite, and those calls run inbound into the platform.

SourceGingerControl, gingercontrol.com/faqread 2026-09-07

The trade compliance agent handles classification, monitoring, reclassification and alerts autonomously. The audit layer reruns the reasoning on duty decisions, reconciles purchase orders to entries and audits the charges in customs value. GingerControl runs this multi step domain pipeline itself. Customers do not compose their own flows, the work does not run across multiple agents, and no runtime is named.

SourceGingerControl, gingercontrol.com/faqread 2026-09-07

Classification runs on a maintained knowledge structure. A GRI 1 to 6 reasoning engine works over a database of more than 20,000 HTS codes with sub millisecond lookup, updated in real time for executive orders, USTR actions and CBP rulings. Section and Chapter Notes and CROSS rulings are read as active decision inputs, and per client product masters persist in the platform.

SourceGingerControl, gingercontrol.com/faqread 2026-09-07

The AI works with human in the loop checkpoints on every high stakes classification or audit response, and those checkpoints are part of the product. The researcher pauses and asks clarifying questions instead of assuming, and it supports pause and resume while the customer gathers missing detail. Compliance Radar flags policy changes for a person to apply, with one click recalculation. The licensed broker's filing review is the customer's own legal gate, separate from these product checkpoints.

SourceGingerControl, gingercontrol.com/trust and gingercontrol.com/faqread 2026-09-07

Security rests on an ISO 27001 aligned ISMS and SOC 2 Type II ready operations, with a report shareable under NDA. GingerControl is aligned and ready, not attested. Granular RBAC, hierarchical permissions, mandatory MFA and IP allowlisting govern GingerControl's own personnel and production systems, not the customer's organization. Customers can configure retention and opt out per organization and scope features and data by team. There is no SSO and no customer role model. Data is encrypted, the platform follows a Zero Trust posture, and penetration tests run annually.

SourceGingerControl, gingercontrol.com/trustread 2026-09-07

Every classification produces a full reasoning chain by default. It includes GRI citations, Section and Chapter Notes, CROSS ruling references and a staged determination at the 4, 6, 8 and 10 digit level, with confidence scores and timestamps. Change logs are preserved for evidence. The record shows why the engine landed where it did, and it is built to support CBP reasonable care review and CF 28 response.

SourceGingerControl, gingercontrol.com/faq and gingercontrol.com/trustread 2026-09-07

A classification case can be paused and resumed. The user pauses, gathers missing information from the manufacturer, and resumes the same case without restarting, with progress and reasoning history fully preserved. That working state persists per case beyond a single sitting. There is no separate memory layer, and nothing says how long saved state lasts or how a customer would find and delete it. The per client HTS catalogs are the application's data model, and the selection history is the audit record.

SourceGingerControl, gingercontrol.com/faqread 2026-09-07

Production workloads can be pinned to regions that match the customer's data residency requirements. Enterprise customers can request dedicated environments and custom retention. There is no published region list, and customers cannot pick a region themselves. The dedicated environment is not a standard customer environment option and comes only on request. The private VPCs are GingerControl's own infrastructure.

SourceGingerControl, gingercontrol.com/trust and gingercontrol.com/faqread 2026-09-07

The Live AI Compliance Hub carries a set of named instruments a customer picks from and uses immediately, with no sign up. They are the HTS Classification Researcher, the Tariff Calculator, the Trade Policy Digest and ECCN export classification, with Compliance Radar and the Product Sandbox alongside. Each works on its own. The Calculator computes the full tariff stack without the Researcher, and the Digest briefs without either. Customers adopt these instruments as they are, and there is no browsable catalog.

SourceGingerControl, gingercontrol.com/faqread 2026-09-07

Policy changes wake the agents. Regulatory monitoring runs 24/7 and fires automatic reclassification alerts when tariff rules or the HTS schedule change and affect the customer's products. The Trade Policy Digest delivers daily alerts on rate changes. Classification itself is started by a user or arrives through the API. Customers cannot set their own schedules, webhooks or inbound queues.

SourceGingerControl, gingercontrol.com/faqread 2026-09-07

A purpose built GRI reasoning engine and a multi angle essential character engine, both GingerControl's own, do the work. No model provider is named, and neither customers nor admins can choose a model or routing.

SourceGingerControl, gingercontrol.com/faqread 2026-09-07

A REST API serves GingerControl's own platform, with a readable reference page and curl examples against api.gingercontrol.com. A single product endpoint returns the HTS code and full tariff stack. A batch endpoint takes up to 200 items and completes in three to five minutes, and component splitting handles Chapter 91 goods. Production keys are sized to calling patterns, with IP allowlists and peak QPS, and uptime is SLA backed. The API has processed more than 100,000 daily classifications in production. Outside callers can drive the platform entirely through it.

SourceGingerControl, gingercontrol.com/products/openapi and gingercontrol.com/products/api-introread 2026-09-07

The only evaluation is GingerControl's own. An internal benchmark of more than 1,000 products reports 99.86 percent accuracy, a production accuracy figure is given, and the team pressure tests the classification logic as a release practice. Confidence scores attach to each output, not to a change under test. The audit infrastructure under services is the Audit System Build engagement, not the platform. Customers get no test harness, scored test cases, release gate or optimization loop to run or buy as a layer.

SourceGingerControl, gingercontrol.com/products/openapi and gingercontrol.com/faqread 2026-09-07

The agents classify from uploaded documents (PDF, JPG, XLSX) and monitor regulatory sources. That monitoring builds and refreshes the tariff corpus the work later queries. The agents do not operate a browser, desktop or computer at run time.

SourceGingerControl, gingercontrol.com/faqread 2026-09-07