Control is structural,
not a module.
A platform that releases payments is a control platform or it is useless. That is why the control level does not sit beside the modules but underneath all of them: identity, permissions, segregation of duties, limits, audit trail and data quality apply the same way everywhere.
How that shows up in operation.
Every number has a context.
Entity, period, source and the transaction it arose from.
Every exception has an owner.
Not an email without an addressee, but a case with a name and a date.
Every decision has an authority.
Checked at the moment of the decision, per entity, amount and transaction type. An approval for step A is not an approval for step B.
Every action leaves evidence.
Origin, basis, decision and outcome stay retrievable together.
Every control has a state.
An internal control system does not live on its description but on its running state: executed, evidenced, open — per period and entity.
Every context has a boundary.
Sensitive records stay within the circle that needs to know. Whoever joins, joins traceably.
Six elements, the same for every module.
The control levelThe control level is part of the architecture, not a module. A second module brings no second control logic with it.
The payment approval, watched in operation.
This is the point where we can evidence the most — which is why it sits here rather than in a list.
The server-side approval control was observed running in a controlled environment, by a party independent of the implementation.
What we do not claim today.
This list is not small print. It is the reason the rest of this page can be trusted.
Immutability of the audit trail
The audit trail is tenant-scoped and append-only in operation. Immutability is an open requirement in our own delivery — we mark it rather than promise it.
Certifications
We claim no certification and no proximity to one. If your review process needs evidence, we tell you what exists today and what does not.
Bank transmission
The approval is evidenced control. Transmission to the bank and the status response are a target process and are not presented as present.
Retention periods
Retention and procedural questions are yours to settle with your auditor. We provide the mechanism, not the legal assurance.
Three questions from the first conversation.
How is this different from the permissions in our ERP?
It complements them. Permissions in the posting system govern who may post there. This level governs who may decide on a transaction that touches several systems — and keeps the decision on the transaction.
What happens when no rule applies unambiguously?
A case is created and no approval. That is the default and not the exception: where there is doubt, nothing is approved.
Can we configure the approval paths ourselves?
Yes, per tenant. The matrix is validated server-side against the tenant's roles — not filtered in the interface.
How does this relate to our internal control system?
It supplies its running state. An internal control system describes which controls should apply; this level shows which ones were executed this period, what evidence they carry and which exception is open. The chain for that is on internal controls.
Can you judge whether a control is effective?
We carry the mechanics for it as a defined capability: sufficiency of evidence, a judgement with a basis, a risk register. What we can evidence today is the server-side check of the approval. An audit opinion is your auditor's to give, not a system's.
What happens when a decision turns out to be wrong?
It is not overwritten but corrected: the new decision stands beside the old one, with a reason and a timestamp. Reversibility is an operating principle — a transaction you cannot take back without losing the trail is not a controlled transaction.
Thirty minutes on one of your own approval paths.
If your request comes from a security or audit context, send us your questionnaire. We answer yours rather than sending ours, and we state clearly what we can evidence.