About

A core worth defending under examination.

Corlayr is a white label core banking system for electronic money institutions, payment institutions and banks. 49 modules on one book of record, running in your infrastructure, under your brand. It exists because the institutions that most need to be able to prove what happened are usually the ones least able to.

What we set out to fix.

Three things that are true of almost every institution of this size, and that nobody chose.

The ledger is somebody else's

Most institutions this size rent their core. The book of record lives in a vendor's tenant, the data model is whatever the vendor decided, and the answer to a regulator's question depends on a support ticket.

The join is the weak point

A ledger here, onboarding there, screening at a vendor, reconciliation in a spreadsheet. Every join between them is a place two numbers can disagree, and every disagreement is yours to explain rather than theirs.

The brand is not yours

The screens a customer sees carry somebody else's design language, and the ones your officers use carry a different one again. Changing either means a change request and a quarter.

What we are for.

An institution should own its book of record. Not licence a view of it, not reconcile against it, own it: the database in its own infrastructure, in the jurisdiction it chose, with the invariants enforced where they cannot be worked around and an audit trail that cannot be rewritten.

That is the whole of it. Everything else here, the module boundaries, the append-only evidence, the controls that refuse rather than warn, follows from taking that seriously rather than saying it.

Four things decided early.

They are the reason some of this took longer than it looks like it should have, and the reason the parts a buyer tests hold up.

Correctness belongs in the schema

An invariant in application code is an invariant until somebody writes a script. Putting it in the database means the rule outlives the service that was enforcing it, the migration that bypassed it and the engineer who understood it.

A record you can edit is not evidence

Append-only is inconvenient. It is also the only version of an audit trail that answers the question an examiner is actually asking, which is not what the data says now but what it said then.

The awkward states are the product

Any system handles a payment that works. What separates them is the one held for review, the control that timed out, the ownership that resolved to nobody. That is where the work goes.

The demo is the product

Every vendor in this category has a demo environment that does more than production does. Ours is production code with seeded data, which is why we are happy to hand over the console and invite you to try to break the ledger.

Where it goes next.

The system is built around one book of record, so the roadmap is about what else can be brought onto it rather than what else can be bolted beside it. Card acquiring, digital assets and inter-entity settlement are modules on the same ledger, not adjacent products with a reconciliation between them.

49
Modules, 31 in every deployment
29
Ledger tables, invariants in the schema
114
API endpoints, two surfaces
6
Institution types served

Come and pick holes in it.

The most useful conversations start with somebody trying to break the ledger on a call. Bring your licence conditions and your own transaction set.