Two systems that were never meant to meet.
Your POS, your accounting package, your warehouse software and the supplier portal that still emails a CSV at midnight. Integration is where most projects actually go wrong, so it is where we spend week one.
- Week one
- every system mapped first
- Idempotent
- retries that cannot double-post
- Reconciliation
- built in, not bolted on
Who this is for
You probably need this if…
01
The same number is different in two systems
Your POS says one thing, your accounting package says another, and reconciling the two of them is somebody's entire Monday morning.
02
A person is the integration
Someone exports from one system and imports into another. They are completely reliable right up until the week they take leave.
03
A vendor says it integrates
It does, in a demo, with clean data. Under your volume and with your edge cases the claim turns out to be more complicated.
What this covers
The work, in detail.
7 capabilities
01
POS & retail platforms
Square, Toast, Clover, Lightspeed, and the older terminals that will only speak to something on the local network.
02
Payment routing
Gateways, processors and the reconciliation layer that proves the money arrived where the order said that it would.
03
Identity federation
SSO, SAML, OIDC and SCIM provisioning, so that someone who leaves loses access the same day rather than eventually.
04
Third-party orchestration
Rate limits, retries, backoff and the queue that stops a partner's bad afternoon from quietly becoming your bad afternoon.
05
Legacy modernisation
A layer in front of the old system so it can be replaced one piece at a time, instead of in a single frightening weekend.
06
Webhooks & events
Signed, replayable, idempotent delivery, with a dead-letter queue that a human being can actually open and read.
07
File-based interchange
SFTP, EDI and the nightly CSV a supplier will never stop sending. Parsed, validated, and alerted on when the format quietly changes.
How we approach it
Positions we actually hold.
Opinions cost something to have. These are the ones we would argue for on your project, including where they make the work slower.
01
Map before you write
Week one is a written inventory of every system, who owns it, what it can emit and what it refuses to. Integrations fail on assumptions, not on code.
02
Assume delivery happens twice
Every write is idempotent. Networks retry and partners replay, and a payment that posts twice is a far worse day than one that fails loudly.
03
Reconcile continuously
A scheduled job compares both sides and reports the drift. Finding a mismatch in an alert on Tuesday beats finding it in an audit in April.
04
Fail loudly, degrade quietly
Integration errors page somebody. The customer-facing side queues and retries rather than showing anyone a stack trace.
Technology
What we work with.
Defaults, not requirements. If you already run something else and have a team who knows it, we work in yours.
- Commerce
- Square, Toast, Clover, Shopify, Lightspeed
- Payments
- Stripe, Adyen, Authorize.net
- Identity
- Auth0, Okta, Entra ID, SAML & OIDC
- Transport
- Webhooks, SFTP, message queues, EDI
Where this shows up in delivery
- OneHubPOS connects point-of-sale terminals to ecommerce stock in both directions, including the reconciliation pass.
- Our inventory products integrate supplier feeds arriving as anything from a REST API to a nightly CSV.
- Sill ships signed, replayable webhook delivery with an inspectable dead-letter queue as standard.
Bought as part of
Ecommerce & POS
Fixed-price project
Systems Integration
Fixed-price project
Legacy Modernisation
Fixed-price project
This practice is never sold on its own. It is quoted inside one of the engagements above, as part of a single number.
Proof
Where we have actually done this.
Projects we designed, shipped and wrote up. Each one names the decision that was genuinely hard to get right.
Questions
What people ask about systems integration.
- The vendor says there is no API.
- Sometimes that is true, and sometimes it means there is no public documentation. We check the network traffic, the partner programme and the database. If there is genuinely no route in, you hear that in week one rather than month two.
- Can you integrate without replacing our current system?
- That is usually the right answer. A layer in front lets the old system keep running while new work is built against a clean interface, and it turns a risky replacement into a gradual one.
- What happens when a partner changes their format?
- Every inbound payload is validated, and a failure raises an alert. You hear about it from us before you hear about it from a customer.
- Who owns the credentials?
- You do. They live in your secret manager. We request scoped access for the work and hand it back at the end of the engagement.
Related reading
Written on this, by us.
Data & Integration · 7 min read
The Same Number in Two Places
Your POS says one thing and your accounting package says another. Reconciling them is somebody's entire Monday. Here is why systems drift apart, and the boring job that keeps them together.
How We Work · 8 min read
How to Replace a Legacy System Without a Bad Weekend
Big-bang rewrites fail at a famous rate. The alternative is slower on paper, far more likely to finish, and keeps the business trading the whole way through.
Data & Integration · 6 min read
Idempotency: The One Word That Stops You Charging Someone Twice
The most common serious bug in payment code is not fraud or a failed gateway. It is a retry that creates a second charge, and it is entirely preventable.
Tell us what you’re trying to build.
Describe the problem in your own words. We’ll come back within one business day with a scope, a number and a date.
Phone
+1 (407) 796-2376Reply time
One business day, from an engineer
Based in
Orlando, Florida · serving the United States
What happens next
- A reply within one business day, from an engineer
- A thirty-minute call, with no qualifying call before it
- A written scope and a fixed number, if it fits