AI-native operations engineering · Fayetteville, North Carolina · Founder-led
Centralux takes an operation that currently lives in spreadsheets, texts, paper tickets and four disconnected apps, and rebuilds it as one instrument you can read: estimate, approval, schedule, invoice, payment — every transition recorded, every automated action traceable to its source.
An operation that has grown past about ten people has usually accumulated one tool per problem. Each one works. Together they do not add up to an operation you can see. The estimate lives in one place, the approval in a text thread, the schedule on a whiteboard, the invoice in accounting, and the answer to “did this job actually make money?” lives nowhere at all — it has to be reassembled by hand, by the one person who remembers.
The cost is not the software subscriptions. It is the re-keying, the jobs that get scheduled before they were approved, the invoice that goes out three weeks late because nobody owned the handoff, and the quiet decisions made on numbers nobody can trace back to a source.
One system of record
Every job, quote, approval, invoice and payment carries one identity and one history. Questions get answered by query, not by reconstruction.
Four channels. Most engagements start with one and add the next once the first is running in production. Full detail, including inputs, outputs, interfaces and handover, is on the services specification.
Your scattered tools, spreadsheets and threads consolidated into one system of record with a single identity per job, customer and document.
Estimate, approval, schedule, invoice and payment wired into one flow, with an audit trail on every transition and a human gate before anything leaves the building.
Ask your own business questions in plain language and get an answer with the query and the source rows attached, so you can check the work instead of trusting it.
The foundation under the other three: isolation, provenance, reversibility and failure behaviour designed on purpose rather than discovered during an incident.
These are the tolerances we hold on every build. They are the reason to hire an engineering firm rather than a strategy firm, and they are the first thing we will put in writing.
A specification is only useful if it says where the instrument stops working. Ours stops here.
We do not sell an AI readiness assessment, a maturity model or a roadmap document. The deliverable is a running system you use on Monday.
Dropping an assistant on top of disorganized data produces confident wrong answers. The data work comes first; the interface is the last mile, not the product.
We are not placing contractors into your org chart by the hour. Engagements are scoped work with a defined output and a defined end.
We do not train foundation models or publish papers. We compose proven components carefully, which is a different and less glamorous job.
If your accounting software works, it stays. We integrate around what is load-bearing and replace only what is actively costing you.
Below roughly ten people, an off-the-shelf tool is usually the right answer and we will say so on the call. We are the right call when the tools have stopped fitting.
We map the flow you actually run — not the one on the org chart — and write down where records are created, where they are re-keyed, and where they go missing. The output is a written specification with a scope, a fixed price and an explicit list of what is out of scope.
The specification gets built and put in front of the people who will use it, in production, on real work. We hold the operating conditions in § 03 throughout. You get the source, the schema and the runbook at the end whether or not there is a Phase 02.
Two honest options. Either we stay on a defined monthly scope for changes, monitoring and the next channel — or we hand the whole thing to your team or your existing IT provider and walk. Both are quoted; neither is a trap. There is no auto-renewing term.
The full engagement model, including how we price and what we need from you, is on the about page.
Bring the flow that is costing you the most. We will walk it end to end, tell you which parts are worth automating and which are not, and if the honest answer is that you do not need us yet, you will hear that instead of a proposal.