QgenticQgentic
Qgentic / how it works / integration
Integration · for compliance and IT

ERP integration that only ever reads.

Qgentic reads from your ERP and core systems through read-only feeds in standard formats, with no proprietary module and no integration programme. There is no write path, so your systems of record stay exactly as they were.

Three feeds, all read-only

Everything comes in through three standard, read-only feeds. Validation, assembly and export are decoupled from the source, so an ERP migration never touches your compliance logic.

FeedWhat it carries
list documentsAn index of the contracts and agreements available to process.
read documentThe full text of a given document.
master dataEntity information, function catalogues and vendor metadata (names, LEIs, locations, contacts).
Validation

Validated at the boundary

Every record is validated against a strict schema before processing. A record that fails is flagged with the exact path to the broken field (for example vendors.3.lei: invalid checksum) and kept out of the register, so reporting starts from validated records only.

The integration surface: read-only by construction ERP and core systems, middleware and drop zones feed four connectors (rest, sftp, filesystem and csv), all pull-only against one boundary contract. Every record passes a schema gate that validates identifiers against their standards and refuses non-compliant records with a precise error path. The engine's validation, assembly, approval and export are decoupled from the source, every ingested byte is hashed onto the audit chain, and the write path back to the ERP does not exist: the connector interface has no write, update or delete call. § YOUR ESTATE § THE CONNECTORS § THE BOUNDARY GATE § THE ENGINE ERP & core systems SAP · Oracle · Dynamics NetSuite · Sage · GRC Your middleware SAP CPI · MuleSoft · Boomi or none: direct feeds Drop zones SFTP · share · bucket read-only mounts Four connectors, one contract rest pull · scoped bearer token sftp scheduled pull · no inbound port filesystem watched directory · stateless csv RFC 4180 batch exports PULL ONLY · NOTHING WRITES BACK The connector contract contains no write, update or delete operation. no write path into your systems of record SCHEMA GATE Every record validated before processing. vendors.3.lei: invalid checksum refused, with the path to the broken field ISO 17442 · 3166 · 4217 identifiers checked against their published standards The engine Validation, assembly, approval and export, decoupled from the source. every ingest hashed SHA-256 audit chain from the first byte read An ERP migration never touches compliance logic. new feed, same record THE WRITE PATH DOES NOT EXIST no write, update or delete call exists in the interface; read-only is architecture, not a permission setting Read-only by construction: four connectors, one boundary contract, every ingested byte validated and hashed, and no code path that could write to your ERP.
The integration surface. Four connectors cover every ERP tier (direct REST, middleware drops, SFTP pulls, batch CSV), all pull-only against one boundary contract. Every record passes the schema gate before the engine sees it, with identifiers checked against their standards and failures refused with a precise error path. The write path back to your ERP does not exist.

Three ways in, by ERP tier

Pick the route that suits your ERP. None of them needs an ERP compliance module.

Tier 1 · Direct API

Native APIs

Connects to native ERP APIs or reporting modules, such as SAP OData/BTP, Oracle Fusion REST, Dynamics 365 OData or NetSuite SuiteTalk, through the built-in rest connector, optionally via an iPaaS layer.

Tier 2 · Middleware

Automated file transfer

Your ERP's standard reporting generates the files (SAP AL11 exports, for example). Your integration layer (MuleSoft, Boomi, SAP CPI) moves them to a drop zone, and the filesystem or csv connector takes it from there.

Tier 3 · Flat files

Batch CSV exports

Scheduled CSV exports, common in mid-market systems such as Sage or NetSuite saved searches, collected automatically from a secure drop directory or SFTP share.

Every tier moves data system to system, with no re-keying, and every ingestion event is on the audit chain.

The four connectors

Each connector is configured with environment variables, to fit the extraction you already run.

The filesystem connector

Watches a directory for new data, such as a read-only cloud bucket or a network share, and re-reads it statelessly on each run.

The csv connector

Parses standard flat-file batch exports. It needs a header row, supports RFC-4180 quoting and takes the columns in any order.

The rest connector

Reads three standard JSON API endpoints with bearer-token authentication. TLS termination, IP restriction and token management stay at your API gateway.

The sftp connector

Pulls on a schedule from your SFTP server, so no inbound firewall change is needed. Credentials are stored encrypted.

custom interfaces

A proprietary source (a message queue, a database view, an SDK) plugs in by implementing the same read-only interface.

QGENTIC_CONNECTOR=rest QGENTIC_ERP_BASE_URL=https://gateway.example.com/api/qgentic

Files as your ERP names them

Qgentic takes the files your ERP already produces, named and formatted the way it already does it, so nothing upstream changes.

  1. Pattern matchingEach source is a file pattern such as UMSV_*.csv, which fits the ERP's default naming.
  2. Latest fileThe most recent file matching the pattern is the one processed, so the directory needs no housekeeping.
  3. NormalisationBOMs, currency symbols and varying delimiters are handled on the way in.
  4. Computed figuresDerived figures and totals are computed inside the application, which keeps the digital link intact.

Field mappings for the main ERPs

Reference mappings for the main ERP systems. Fields from GRC or operational resilience tools merge in through the same standard feeds before ingestion.

Target field SAP S/4HANA Oracle Fusion Dynamics 365 NetSuite
vendor.nameBUT000 / LFA1.NAME1Supplier NameVendorsV2.VendorOrganizationNamecompanyName
vendor.leiBP identification (LEI type)D-U-N-S / LEI idRegistration IDs (LEI)legalEntityId
vendor.countryLFA1.LAND1Supplier Address CountryAddressCountryRegionIdVendor Address Country
vendor.contactEmailBP Contact EmailSupplier ContactEmailemail
DocumentsCV01N, AribaProcurement AttachmentsSourcing AgreementsFile Cabinet Attachments

Every field is validated at the boundary: LEIs against ISO 17442, countries against ISO 3166, currencies against ISO 4217. An invalid record raises an explicit failure.

Security controls

Read-only access

Every connector is read-only: the connector interface has no write, update or delete operation to call against a source ERP.

Schema gates

Data is validated against its schema before processing, and a malformed input is logged with a precise error and kept out.

Transport

REST APIs authenticate with bearer tokens and enforce timeouts. TLS and access control sit at the edge gateway. SFTP runs on managed credentials.

Data residency

Processed data, compliance registers and final reports stay inside the deployment boundary, and air-gapped configurations actively block egress.

Audit logging

Every action, including ingestion, transformation and export, is logged to a tamper-evident SHA-256 hash chain that an auditor can verify independently.

Where it runs

Stateless services that run in the multi-tenant cloud, in a dedicated environment, or fully air-gapped on your own infrastructure.

From mapping to production

Integration is fixed in scope and runs in four steps.

  1. MappingA workshop with your team settles the data mappings for your ERP configuration and reporting needs.
  2. Build and testThe feeds are set up and validated against the interface contracts in a staging environment.
  3. Parallel runQgentic runs in shadow mode beside your current process, so its output is checked before anything depends on it.
  4. Go liveIt moves to production, with the approval workflow and cryptographic auditing switched on.