Custom
Integrations
The integration layer as the project rather than as a task: several systems that were never designed to agree, one version of the data, and failure treated as a normal condition rather than an exception.
The problem
Every system in the company is the source of truth for something and wrong about everything else. The order in the CRM, the invoice in accounting and the stock in the warehouse describe the same day differently, and reconciling them is somebody's week.
Why a package will not do
Connectors are sold as a solved problem. They are, for the case where two modern systems already agree about what a customer is. Most companies are not that case: the systems arrived a decade apart, one has no API worth the name, another is a reporting endpoint you do not get to negotiate with, and none of them define a customer the same way.
The hard part was never the transport. It is deciding which system owns which fact, what happens to a record that exists on one side and not the other, and how a correction made in the wrong place propagates. Those are business decisions wearing technical clothes, and no connector takes them for you.
Then there is the half nobody demonstrates: what the integration does when a system is down, when a response is late, when the same message arrives twice, and when a figure has to be reconciled rather than overwritten. Retries, idempotency and a record of what moved — so a disagreement is settled by looking rather than by arguing.
State systems are a category of their own. e-Factura, ANAF and SPV are not a setting to switch on: they are an obligation with a schema, a calendar and consequences, and they belong in the architecture rather than bolted to the end of it.
How we approach it
01
Discovery
Which system owns which fact today, where the same record exists twice, and which disagreements are already costing somebody a morning.
02
Architecture
The direction data flows, who wins a conflict, what is synchronous and what is not, and how failure is expected to behave.
03
Build
Built with retries, idempotency and a record of what moved, then run against real volume before anything depends on it.
04
Scale
New systems, changed schemas and the volume that arrives later, added without redoing what is already connected.
Proof
Econet Auto, an engine reconditioning plant. Six supplier catalogues arrive as separate services, and goods receipt posts to the Romanian chart of accounts.
Is this the right fit?
For companies where several systems have to agree and currently do not, and where being wrong has a cost somebody can name. If what you need is one or two handoffs between systems that already have proper APIs, the automation offer on the Solutions line is faster and cheaper.
It starts with a discovery: which system owns which fact, written down, before anybody agrees a price for connecting them.
What a discovery includesContact