The government practice
Ask your registry a question. Get an answer you can prove.
We survey a records estate, extract it from paper and scan, and build retrieval where every answer names the file, the folio and the date, and links to the document itself. Then we stay and operate it. Hosted in Nigeria, because government data has to be.
The problem
The records exist. Nobody can ask them anything.
A question that should take a minute takes three weeks, and at the end of it nobody can prove the answer. That gap is the work, and it is what we do.
Not a data warehouse. Not an API. In most cases: paper files in a registry indexed by file reference and folio, scans from a digitisation project that stopped halfway and has no index, an Excel register whose authoritative copy is on somebody’s laptop, a legacy system holding one period’s records, a portal a previous vendor built that nobody opens, and WhatsApp carrying the actual coordination.
We design for that, not for the ministry anyone wishes existed. The system’s first job is to make the estate answerable without demanding the ministry reorganise itself first.
Where we start
Records Readiness Review
In a month, an honest inventory of what your registry actually holds, and proof of what it would take to make it answerable.
We inventory one records estate: what exists, in what medium, in whose custody, at what quality, in what volume. Then we test extraction on a real sample of your actual documents, not on assumptions. You leave with a records inventory, a feasibility assessment, a costed roadmap, and a working proof. We will not quote a build without this, because extraction is where these projects die and where vendors under-quote.
3–5 wks · fixed fee · naira
Deliberately small and self-contained, and scoped to sit on a line your vote already carries, usually records digitisation or software acquisition.
And then
Climb as it proves out.
Build & Ship
We take it to production, and own the result.
We take the highest-value use case to production: extraction, indexing, retrieval with citations, security, monitoring, and integration with what you already run. Milestones map to deliverables you can see, and for government work to certificates you can sign.
Run It For You
We operate it after handover, and hand it over properly if you ever want us gone.
We operate the system after handover: monitoring, evaluation, retraining, support. Every engagement also carries an exit package from day one, an export in open formats, the source, a runbook, and a transition period. A client who can leave cheaply stays for the right reason.
Compliant by Design
Your data stays in Nigeria, you own the index, and every answer is auditable.
Government data stays in Nigeria, because NITDA requires it and because it is the right default. Sensitivity classification fails closed. Every answer, routing decision, model version and human review is logged to an append-only trail you can query without asking us. DPIAs, retention aligned to your own schedule, and FOI readiness.
Also
Records are where we go deepest. They are not all we do.
Two things ministries ask us for directly, and both come up in almost every conversation that starts with records.
Product & Interface Design
Built is not the same as used.
A system that officers avoid, or customers abandon halfway through, has failed regardless of how well it was engineered. We design the flows and the interface alongside the build, with the people who will actually use it, and we treat the ugly screens as the important ones: the internal form a registry clerk fills in forty times a day matters more than the landing page.
Modernisation & Migration
Move off the old system without a big-bang rewrite.
Most institutions here are not starting from nothing. They are carrying a legacy system that half works, a vendor platform nobody can extend, and data spread across it in inconsistent shapes. We replace these incrementally: stand the new system beside the old one, move one capability at a time, reconcile the data and prove the reconciliation, and keep both running until the old one is genuinely unused. Big-bang rewrites are how these projects fail, and we will argue against one.
Modernisation is usually the more urgent of the two. Most MDAs are carrying a vendor platform nobody can extend and a portal a previous supplier left behind, and the honest answer is rarely to rewrite everything at once. On design: the screen that decides whether a system gets used is the internal form a clerk fills in forty times a day, not the public landing page. Full service list.
How it works
Survey, extract, index, cite, operate.
Survey
Inventory the estate: what exists, in what medium, in whose custody, at what quality and volume. Extraction tested on a real sample.
Extract
Capture, deskew, OCR, and structure. Government scans are harder than bank documents: photographs taken at an angle, handwritten marginalia, stamps overlapping the text they authenticate, bound volumes scanned across the gutter. This is most of the engineering, and it is where these projects die.
Index
File reference, folio and date preserved on every fragment, so a record stays identifiable rather than dissolving into a searchable blob.
Cite
Every answer names the file, the folio and the date, and links to the source image. When there is no record, the system says so rather than inventing one. Contradictions between registries are surfaced as findings, not silently resolved.
Operate
We run it after handover. And the exit package exists from day one, so you are not trapped the way the last portal trapped you.
Residency
Your records do not leave Nigeria.
NITDA requires government data to stay in-country. That is stricter than the general cross-border regime and it is not waivable by paperwork, so for records work the default is in-region and the exception has to be argued for, rather than the other way round.
Classification fails closed: anything uncertain is treated as restricted. The rules are written down and reviewable by you, not buried in code you cannot inspect, and there is a test suite asserting that record-class data cannot reach an offshore endpoint.
Ownership
You own it. And you can leave.
The ministry owns its records, the derived index, the extracted data, the embeddings, and the audit log. Not licensed. Owned. We do not train on your data, aggregate it, benchmark with it, or reference it in marketing. What we carry to the next ministry is engineering, never data.
At any point, on request, without penalty: a full export in documented open formats, the source for anything bespoke, a runbook sufficient for another competent firm to operate it, and a transition period during which we assist your successor.
Originals are never destroyed, moved, or re-filed unless you direct it in writing. Digitisation supplements the registry. It does not replace it, and it must not disturb the custody chain of the paper.
Limits
What we will not build.
Not subject to negotiation on a per-engagement basis. If work requires any of these, we decline it and say why.
- Biometric identification or face matching against a population.
- Predictive policing, or any system that scores individuals for likelihood of offending.
- Covert monitoring of citizens, employees, or officials.
- Automated decisions on citizen entitlements. A machine must not decide who receives a service, a permit, or a payment.
- Aggregation of citizen data across agencies for general surveillance purposes.
- Resale, secondary use, or model training on citizen data.
- Systems designed to obstruct audit, including anything that suppresses or edits its own log.
Contact
Let's talk about what you are building.
Twenty minutes. We will tell you what we would do, and say so plainly if we are not the right firm for it.