Dexon BPM platform

ETL and integration

Orchestrating what you already have, instead of replacing it.

The most expensive sentence in an enterprise proposal is “and we migrate the rest”. Dexon BPM is built on the opposite premise: the systems of record stay where they are, and the process crosses them.

API-first, and demonstrably so.

Dexon BPM exposes authenticated REST services, and its own web and mobile clients consume them. That matters more than the phrase “we have an API”: the interface is not a side door added for integrators, it is the same surface the product itself runs on — so it is as complete as the product, and it does not fall behind.

Alongside it, the integration engine handles scheduled and event-driven exchange with databases, files and web services: bulk loads, synchronisation, reconciliation and outbound updates, configured rather than coded.

What it does

The ways information gets in and out

Integration is part of the core toolset here, which is a different claim from integration being possible on a project.

REST API

Authenticated services over record types, relationships and history. External automation can create records, update resolution fields and close cases, attaching the originating identifier so the systematic origin stays in the audit trail. Each consuming system gets its own credential, generated by the customer’s administrator and revocable without affecting the rest.

Integration engine →

Scheduled or event-driven flows against databases, files and web services: bulk creation and update, synchronisation with external repositories, and outbound updates to the systems that need to hear about an outcome. Web services are read from their WSDL, OpenAPI or HTML documentation, with pagination; files travel over FTP, FTPS or SFTP, and messages over AMQP queues. Change-capture triggers, query monitors and mapping rules are configured from the web, and integrations are promoted between environments with their log.

Any database engine →

SQL Server, Oracle, MySQL, PostgreSQL and any ODBC or OLE DB source, read and written with each engine’s own native queries — so the integration speaks the dialect of the system it connects to.

Loads built for production →

Every integration is previewed node by node without writing anything, resumes from the chunk where it stopped without repeating finished parts, and splits into threads shared across several process servers. Per-node thresholds — minimum and maximum rows, maximum error rate — stop an empty or corrupt file before it empties or dirties the destination.

Excel, both ways →

Master data is loaded and maintained by downloading an Excel template of the table, with drop-down lists for its references, and importing it back: rows are updated or inserted and a summary is returned. A file from another table cannot touch the records, and an import can open an activity by itself.

Directory and identity

LDAP and Active Directory for authentication, user provisioning and directory lookups, over HTTPS and TLS.

Email, both directions

Inbound mail creates or updates records, capturing content, sender and attachments in the history, with the original message preserved on the record. Outbound notifications are template-driven and remain part of the communication trail.

Collaboration

A conversational assistant, and a Microsoft Teams bot available from November 2026, so an interaction that starts in a chat lands on the record instead of beside it.

Monitoring and discovery →

Native discovery and monitoring feed records directly. External monitoring platforms enter through the API, the email connector or the integration engine, and the originating event reference is retained.

DevOps

Pipeline events can create or update records, relate test and validation results to a change or a release, update component records and open a problem for a promoted defect — through the API and integration patterns.

Data lineage →

Where several sources describe the same object, the configuration declares identity matching, source priority, survivorship and exception handling — and retains which source is authoritative for each attribute.

Legacy and EDI formats

Fixed-structure COBOL records, fixed-width, binary and sequential files, XML, and EDIFACT and X12 messages — reading both, writing EDIFACT — with XSLT transformation and FTP, FTPS or SFTP transport, configured in the implementation.

Platform health for your monitoring →

Your corporate monitoring tools query the platform over SNMP through its own MIB under Dexon’s enterprise number: database availability and latency, active process servers, work queue and oldest task, pending and failed mail, open activities, licence expiry and consumption, inventory and integration runs. The platform also measures its own API — requests per minute, server errors, 95th-percentile latency, background tasks, database space — and watches itself with its own engine: a platform-operation process with nine indicators emails the warnings and opens a ticket on critical breaches.

The hard part

A consolidated view is five decisions, not one connector

Federating two inventories is where most CMDB projects quietly fail. Not because the data cannot be moved, but because nobody decided these five things in advance.

The ruleWhat it decides
Identity matchingWhen two records from two sources are the same real object
Source priorityWhich system wins when they disagree
SurvivorshipWhich value survives the merge, attribute by attribute
Exception handlingWhat happens to the rows that satisfy none of the rules — instead of vanishing
LineageWhich source each surviving value came from, so an assessor can tell an observation from an imported assertion

Precision

The distinctions this module preserves

Pairs of terms used as synonyms that are not — and an implementation depends on the difference.

An observation and an authoritative assertion →Discovery reports what it saw. An imported record asserts what a system of record holds. The consolidated view keeps the difference visible, because reconciling the two is a decision someone has to own.
A protected channel and a protected payload →TLS protects the channel and is terminated at proxies and inspection appliances. Dexon additionally encrypts every payload with AES-256 at the application layer, so the data — not just the connection — stays confidential, and stays so against a future quantum computer.
Orchestration and system of record →Where financial values, currency, tax or ledger postings are authoritative elsewhere, they stay there and are exchanged. The process reads and writes them in their system of record, so it solves reconciliation instead of creating it.

On the roadmap

What arrives next, and when

Each item is available for a demonstration or a proof of concept from the month shown.

November 2026

Web connection editor

Create external connections and configure every integration node from the web.

December 2026

Packaged application connectors

Ready connectors for the most common corporate applications, instead of building each interface from scratch.

The proof

Integration is what makes the rest verifiable

A certified process is only as good as the data reaching it. Directory integration supplies the identities that approvals resolve against; discovery supplies the components that impact analysis reads; inbound email and the API supply the events that open records. The audit question is never “do you have an API” — it is whether the record can show where each piece of information came from.

See the twelve certified processes →

Bring the integration you have been told is impossible

The legacy system with no API, the file that lands at three in the morning, the directory with three sources of truth. We will show you how it connects, on your own systems, before you commit.

Book a technical session