Integrations

Custom integrations between systems that were never built to talk to each other

ERP, e-commerce, marketplaces, APIs and legacy databases. When no ready-made connector exists, we build one — and keep it running with no human intervention, every day.

What we do

Different systems, a single operation

Most companies do not have a software problem. They have a distance problem between pieces of software. The ERP is fine. The online shop is fine. The marketplace is fine. What is missing is the path between them — and that path is precisely the one nobody wants to build, because it means understanding both sides deeply.

That is what we have done for more than fifteen years.

We develop custom integrations: orders that land in the ERP on their own, stock that updates without a spreadsheet, records that are never retyped, finance that arrives reconciled. We work both with well-documented APIs and with closed, ageing systems that have no vendor support at all — the scenario where most suppliers give up, and where, honestly, we do our best work.

Types of integration

What we connect

ERP and e-commerce

Orders, customers, payment methods, personalisation and size-and-colour grids flowing from the online shop straight into the ERP, ready for invoicing and dispatch.

Stock and price synchronisation

Thousands of SKUs updated automatically, taking grid variations, multiple branches and products not yet registered at the destination into account.

REST and SOAP APIs

Consuming and publishing third-party APIs, handling authentication, versioning, rate limits and contract changes.

External channels and data import

Orders and bookings raised on third-party platforms imported, converted and written into the company’s system — translating codes, formats and values between two data models that were never meant to fit together.

Legacy databases

Integration with systems that have no API, no documentation or no support, through a complete mapping of the data model — always with the client’s authorisation and on the company’s own database.

Routine automation

Processes that run by themselves: imports, exports, checks, alerts and scheduled reprocessing.

How we work

Method

  1. Diagnosis

    We understand the operation before the code. Which data needs to travel, in which direction, how often, and what happens when something fails.

  2. Technical mapping

    We survey APIs, tables, relationships, keys, sequences and business rules on both sides. Where there is no documentation, we write it.

  3. Proof of concept

    One real flow, end to end, before any promise. This is where the risks surface — and that is deliberate.

  4. Development

    Building the full integration, with error handling planned from day one.

  5. Acceptance with the real operation

    Testing alongside the people who use the system every day, with real data at real volume.

  6. Monitoring and evolution

    An integration is not a delivery, it is a relationship. Platforms change, ERPs are upgraded, APIs break. We stay with it.

Engineering principles

What separates an integration that works from one that causes trouble

Any programmer can read an API on a good day. What keeps an integration in production for years is what happens on the bad days.

Fault tolerance

The internet drops. The API goes down. The server reboots. Our integrations do not die from that: they wait, retry, and pick themselves back up when the environment returns.

Resuming from the exact point

Long processes record their own progress. An interruption halfway through ten thousand products does not start over — it carries on from where it stopped.

Respect for rate limits

Practically every API has an hourly ceiling. We treat that as a design rule, not a surprise: the load is spread over time so it never triggers a block.

Idempotency

Reprocessing does not duplicate. An order imported twice is still one order.

Traceability

Every event is logged. When someone asks “why did that order not come through?”, there is an answer — with date, time and reason.

Translation between data models

Two systems rarely name the same things the same way. Codes, units, date formats, rounding and value rules have to be converted with judgement — it is where the discrepancies that only show up at month-end are born.

Unattended operation

The end goal is simple: nobody needs to remember the integration exists.

Success story

Mundo do Futebol — fifteen years, four generations of integration

In 2010 the online shop and the ERP did not talk to each other, and every order was retyped. Another developer had already tried and failed. We mapped the ERP database record by record and built the integration that, four platforms and two ERP versions later, is still running.

2010
first year in operation
4
generations of the integration
2
different databases
0
orders retyped since then
Read the full story

Who for

Who we work with

We serve medium and large companies that already run systems and need them to work together.

Our ideal project involves at least one of these:

  • An established ERP that needs to connect to sales channels
  • Enough data volume to make the manual process unworkable
  • A legacy, closed or unsupported system somewhere along the way
  • An integration another supplier already attempted and could not sustain
  • A need for continuous operation, with nobody watching over it

Frequently asked questions

Frequently asked questions

Can a system with no API be integrated?

Yes. It is a substantial part of our work. Where there is no API, the integration is done directly against the database, from a mapping of the system’s data model — always with the client’s authorisation and on the company’s own database.

What if my ERP vendor will not support the integration?

A common scenario, and not a blocker. Our flagship case began exactly that way: no useful documentation and no vendor support. The path is to survey the data model methodically, validating every hypothesis with real entries in the client’s environment.

What happens if the internet drops or my e-commerce API goes down?

Nothing is lost. Our integrations are built with automatic retries and resumption from the point of failure. When the environment returns, processing carries on by itself, with no manual restart and no duplication.

How do you handle API rate limits?

We treat the limit as a design parameter. The load is spread over time, with pacing and automatic waiting, so it never triggers a block for excess calls.

Do you work with products that have size and colour grids?

Yes. Grids, variants and personalisation are handled across the whole flow — order, stock and product records — not as an exception.

How long does an integration take?

It depends on the quality of the available documentation and the complexity of the data model. Projects with a documented API on both sides tend to be quick; projects that require mapping a closed system need an initial investigation phase. In both cases we start with a diagnosis that sets the timeline and scope before development begins.

What if the platform offers no API, only file exports?

That is integration too. We import spreadsheets and exported files, convert codes, formats and values into the destination system’s model, and write the records ready for use — automatically and repeatedly, not as a manual chore.

Do you maintain the integration after delivery?

Yes, and we consider it an essential part of the service. Platforms change, ERPs are upgraded and APIs break. The Mundo do Futebol case went through four rebuilds since 2010 precisely because of that continuity.

Do you serve companies in other states or countries?

Yes. We work remotely and serve clients in Brazil and Paraguay.

Let us map your scenario

Tell us which systems you need to connect. We come back with an initial technical assessment, no strings attached.

Talk to Expert Informática