Software that assumes the lights will go out.

Most retail software is designed somewhere with reliable power and reliable data, then sold somewhere that has neither. Nostra starts from the opposite assumption, and a surprising amount follows from that one decision.

Why this exists

The gap was never features. It was assumptions.

Walk into most shops and you find the same arrangement. A point of sale that works when the connection holds. A notebook for when it does not. A spreadsheet somebody updates on Sundays. An accountant who arrives monthly and reconciles three versions of the same week.

Every one of those is a reasonable response to an unreliable system. The notebook exists because the software failed. The spreadsheet exists because the software could not be trusted to count. The monthly reconciliation exists because nobody believes any of them on their own.

The vendors are not incompetent. They built for a place where the network is a utility rather than a variable, and shipped that assumption into a market where it does not hold. Everything downstream breaks, and the shop owner pays for it in lost takings and Sunday afternoons.

Nostra takes the opposite position. The connection will fail. Power will fail. Staff turn over, devices break, and eventually somebody will try to cover a shortfall. Design for that, and the version that works in a good week comes free.

How we decide

Four rules we argue from.

Every product decision comes back to one of these. They are written down because rules are easier to hold to when they are not just a feeling.

The books must agree with themselves

Stock cannot move without a record written at the same instant. A supplier balance is derived from payments, not typed in. Wherever a number could be entered by hand or calculated from evidence, we calculate it.

Failure must be visible

A failed sync goes to a conflict log instead of disappearing. An offline till shows its queue at all times. The worst outcome is not an error. It is a system that looks fine while being wrong.

History is not editable

Financial records are voided, never deleted, and the void is signed. It costs storage and occasionally annoys people who want a tidy screen. It is worth both.

The strict path must be the easy path

Earliest-expiry-first is automatic, so a cashier cannot get rotation wrong. Adjusting stock is harder than selling it. Where doing the right thing takes effort, people eventually stop doing it.

Boring where it counts. Retail software fails quietly: a total that drifts by a few naira, a stock count nobody trusts, a report that disagrees with the drawer. The defence is unglamorous and it is mostly tests.

497automated tests that must pass before any release ships

12modules over one shared ledger

199permissions enforced on the server

0financial records that can be permanently deleted

How we work with you

Close enough to see the shop.

We would rather be standing behind your counter on a Saturday than presenting slides in an office.

We train on the floor

Cashiers learn during a real trading session. The questions asked at 4pm on a Saturday are the ones worth answering.

We watch the first week

We check that stock, drawer and takings reconcile against your hand count until you trust the system more than the notebook.

We say what is missing

The roadmap lists what we have not built yet, with reasons. You hear it from us, not in month three.

Careers

Come and make the boring parts correct.

A small team in Lagos building for a market we live in. If "the sale must not be recorded unless the stock movement is written in the same transaction" sounds like a good time, we should talk.

What we look for

People who read the failure case first
Comfort saying "I was wrong about that"
Willingness to stand behind a counter and watch
An eye for the number that is quietly wrong

Open roles

No posted vacancies right now. We keep good applications on file and come back to them when a role opens, so it is worth writing if this sounds like your kind of problem.

Write to us anyway →

Come and disagree with us

Tell us what we got wrong.

The most useful walkthroughs are the ones where somebody who has run a shop for fifteen years tells us what we missed. Those conversations change the product.

Book a walkthrough →See what shipped