An open-source ERP with no licence fee — what it covers, what makes it unusually configurable, and who it fits.
iDempiere is an open-source ERP and CRM platform, built in Java on a PostgreSQL or Oracle database, descended from Compiere through ADempiere. It covers finance, purchasing, inventory, sales, manufacturing and CRM. Its distinguishing feature is the application dictionary: windows, fields, validation and business logic are defined as data, so much of what would be development elsewhere is configuration here.
Two things make iDempiere interesting to a mid-sized business. There is no per-user licence, so adding the fiftieth user is a hosting decision rather than a purchase. And the application dictionary means a new field, a new window or a new validation rule is usually configured rather than coded — which changes both the cost and the speed of adapting it to how you actually work.
Tables, windows, fields, tabs and validation rules are stored as data. Changing the UI or adding a field is done through it rather than in source code.
A single instance supports multiple legal entities and organisations with shared or separate master data, which suits groups.
General ledger with accounting schemas, payables, receivables, cash and fixed assets, posting from the operational documents.
Purchasing, inventory with multiple warehouses and locators, sales orders, shipments and invoicing, plus manufacturing through the light manufacturing module.
Extensions are packaged as plugins that deploy alongside the core rather than modifying it, so upgrades stay manageable.
A distributor needs a mandatory field on the sales order capturing a delivery instruction code, validated against a list, defaulted from the customer, and shown on the picking document.
In many ERP systems this is a development task. In iDempiere most of it is Application Dictionary configuration: define the column, add it to the window and tab, attach a reference list for validation, set the default, and add it to the print format. No Java, no build, no deployment — and crucially, nothing that a core upgrade will overwrite, because the dictionary is data rather than code.
The part that does need a plugin is any genuine behaviour: if the code must drive routing logic or push to a carrier integration, that is a model validator in an OSGi plugin. The useful discipline is to establish which layer each requirement belongs in before quoting it — teams that reach for Java first end up maintaining code for things the dictionary would have done, and teams that avoid Java entirely end up with validation logic hidden in callouts where nobody can test it.
| Term | What it means |
|---|---|
| Application Dictionary | The metadata layer defining tables, windows, fields and validation — changeable without Java. |
| OSGi | The module framework that lets plugins be added and updated without altering the core. |
| Callout | Logic that fires as a user changes a field, for defaults and dependent values. |
| Model validator | Logic that fires on save or document action, used for business rules and integrations. |
| Document action | The lifecycle operation on a document — complete, void, reverse — each with defined accounting consequences. |
The quickest way to make every future upgrade painful. Use the plugin mechanism, which exists precisely for this.
Implementation, data migration and training cost the same as for any ERP. Budget them properly.
Accounting schemas and the chart of accounts drive how everything posts. They are hard to change later.
An open-source system with nobody who knows it is a liability. Settle support before go-live, not after.
| Metric | What it tells you | What distorts it |
|---|---|---|
| Changes by layer | How much sits in the dictionary versus plugins versus core. Core should be zero. | |
| Plugin count and size | The real custom surface area you maintain through upgrades. | |
| Core modifications | Should be zero. Anything above that is an upgrade liability with a date attached. | |
| Version lag | How many releases behind current. Growing lag is how open-source systems become legacy. |
We build and support the systems this page describes. Start with a free working proof of concept — no cost, no commitment.