Do you have a project in your
mind? Contact us.
Contact Us
API integration and connectors
API integration
When the ready made connector does not exist
Sometimes the integration you need is not in anyone's marketplace, or the one that exists moves three fields and ignores the rest. Sometimes the system is old enough that its API was written before REST was common.
We build the connection properly: authentication, field mapping, rate limits, retries, and a way to see what happened when something did not arrive.
Building a connector
A one way feed or a two way sync between systems that have no existing link. Most of the difficulty is in the disagreements: which system wins when both have changed the same record, and how a failure halfway through is recovered from.
Working with awkward APIs
Undocumented behaviour, aggressive rate limits, no sandbox to test against, or a vendor whose support queue is measured in weeks. This work is slower and we will say so at the start rather than discover it halfway through.
Tools and systems
What we work with
Protocols
Authentication
Building and testing
Running it
What to expect
How a project runs
Reading the API, then testing it
Documentation and behaviour diverge often enough that we work against the real endpoints early. Undocumented required fields, silent truncation and inconsistent date formats all turn up at this stage rather than after the build.
Authentication and access
Getting credentials issued, working out token lifetimes and refresh behaviour, and deciding where secrets live. On enterprise systems this often takes longer than the code, because it depends on someone else granting access.
Mapping the data model
The two systems will disagree about what a customer, an order or a product is. We write down how each field maps, what is required on each side, and what happens to the fields that have no equivalent.
Rate limits, pagination and volume
How much data moves, how fast the API allows it, and what the first full sync looks like as opposed to the daily one. Getting this wrong is the usual reason an integration works in testing and fails in production.
Error handling
Timeouts, partial failures, and responses that return a success code with an error in the body. We make the calls safe to repeat and decide what retries on its own against what needs a person.
Monitoring
Alerting when a sync fails or stops running, and a record of what moved and when, so a discrepancy can be traced rather than argued about.
Handover
The field mapping written down, the credentials documented, and a runbook covering how to re-run a failed sync and what to check first.
Scope
When you should not hire us
- If both systems are on a platform that already links them, and the standard connector moves everything you need, use it.
- If you only need a file moved once a night and both ends can handle a CSV, that is a scheduled job rather than an integration project.
- We do not scrape systems that have a documented API, and we will not build against an undocumented internal endpoint that a vendor can change without warning.
Where enquiries usually go next