Discovery and mapping
Walking your process where it happens, with the people who actually run it. Ride-alongs and shop visits beat interviews every time.
AboutThe company
A small engineering shop
that ships operations systems.
Centralux is an AI-native services company in Fayetteville, North Carolina. We build the systems that hold a working company together — intake, estimates, approvals, scheduling, invoicing, payment — for businesses whose work happens in trucks and on job sites, not in a browser tab.
01What Centralux is
Founder-led. Engineering-first.
Which means: the person you talk to on the discovery call is the person who writes the code and stands behind it. There is no account manager, no handoff to an offshore build team, and no gap between what was promised on the call and what gets built.
Production systems are live for field-service, retail and e-commerce operators. We don’t publish client names or logos — the operators we work with generally consider how their business runs to be their business.
The name. Central plus lux — Latin for light — plus UX. Light at the center: one place where the state of the business is legible, instead of six places where it is partially true.
- Legal entity
- Centralux AI LLC
- Jurisdiction
- North Carolina, United States
- Office
- 109 Hay St Suite 202
Fayetteville, NC 28301 - Structure
- Founder-led, engineering-first. No sales team, no subcontracted build shop.
- Focus
- Field-service and trades operations — the quote-to-cash path in particular
- Engagements
- Scoped phases, flat fees, no lock-in
- Contact
- contact@centralux.ai
02Where we operate
Fayetteville, and the road out of it.
Downtown, at 109 Hay Street.
Our office is in The Hub on Hay Street in downtown Fayetteville. It is a deliberate choice: the businesses we build for are here — the contractors, the service companies, the family operations doing real volume with a whiteboard and a phone. Being local means discovery happens in your shop, on your schedule, looking at your actual paperwork instead of a screen-share of it.
The work itself travels. Systems are built and operated remotely, so an engagement isn’t limited to a drive radius — we work with operators across North Carolina and the wider Southeast, and remotely beyond it when the fit is right. What doesn’t travel well is the first conversation, so when you’re within reach we’d rather have that one in person.
Build and operate
Development, deployment and ongoing operation. Scheduled check-ins, and a written record of every change to the system.
03How we work
Five rules we don’t break.
These aren’t values on a wall. They’re the reasons projects here don’t turn into six-month archaeology digs.
Rule 01
Measure before building
We don’t take a guess about where your time goes. We follow real jobs — including a bad one — and write down what actually happened, with timestamps. The measurement decides the build order.
Rule 02
Smallest useful slice, into production
Not a prototype, not a pilot environment. A real slice of the real path, live, used by real people on real jobs — because a demo tells you nothing about whether it survives a Tuesday.
Rule 03
The old way keeps running
Nothing gets switched off until the new path has proven itself on live work. You should never be in a position where our system failing means your week stops.
Rule 04
Failure is loud
A missing price stops the step and raises a flag. It does not quietly become a zero, and it does not get filled in with something plausible. Silent wrongness is the expensive kind.
Rule 05
Everything is written down
What was built, why, what it connects to, how to run it, and how to take it over. Written for the person who inherits it — including the version of that person who is not us.
04The data stance
Why data integrity is the product.
The pitch everybody makes is “AI will save you time.” The question nobody answers is what happens when it is confidently wrong about an invoice. That question is the entire job.
A system that touches money earns trust in exactly one way: by being auditable. Not by being impressive on a demo, and not by being right most of the time. You have to be able to open it up and see what it did, from what input, at what time, and on whose authority — and you have to be able to do that a year later, when the customer is disputing something and the memory of it is gone.
So we build the record first and the automation on top of it. It is slower to start and it is the only thing that makes the rest of it safe.
Isolation
Single-tenant by default. Your data lives in your own store, under your own credentials wherever possible. No shared table with another client’s jobs in it, and no path by which one client’s data can reach another’s system.
Provenance
Every record carries where it came from: which source, which step, which time, which person or process. A number without a source is treated as an unknown, not as a number.
Gated automation
Anything that sends, bills, schedules or moves money runs on verified inputs. If an input is missing or contradicts another system, the step halts and surfaces instead of proceeding on an assumption.
Your keys, your exit
Credentials live in a secret store, scoped to what they need. The exit path is written down at the start of the engagement: what you get, in what format, and how long it takes.
On models and your data: your operational data is used to run your system and nothing else. It is not pooled across clients, not used to improve anything we sell to anyone else, and not sent to any service the engagement hasn’t explicitly named.
05The engagement model
Phases, flat fees, no lock-in.
Priced per phase and agreed in writing before work starts. No hourly billing, no per-seat licenses, no minimum term.
Phase 0
Discovery call — free
Thirty minutes on how your work actually moves. You leave with a straight answer about whether there’s anything here worth building.
Phase 1
The map — flat fee
A written map of your quote-to-cash path: every step, every handoff, where it stalls, what the delay costs, and a recommended build order. It is a deliverable, not a sales document — it is yours to keep and to hand to somebody else if you’d rather they build it.
Phase 2
The build — flat fee per slice
Scoped slice by slice, each one priced and agreed before it starts. You approve each slice individually, which means you can stop at any boundary without owing for work that isn’t done.
Phase 3
Handover or ongoing — your call
Take the keys and run it yourself, or keep us on month to month for operation, monitoring and extension. Either way, ownership of the system and the data is yours from day one.
What we ask of you: one decision-maker who can say yes, access to the people who actually do the work, and honesty about the parts that embarrass you. The embarrassing parts are usually where the money is.
06Fit
This tends to work when
- You run real volume — enough jobs that the admin has become a second job
- The owner or a key person is the integration layer between the tools
- Work already gets done well; it’s the paperwork around it that leaks
- Somebody can make a decision without a committee
- You’d rather own a system than rent five subscriptions that don’t talk
This tends not to work when
- The real problem is a process nobody follows — software will just make that faster
- You want a proof of concept for a board rather than a system for a business
- Nobody internally is willing to change how a step gets done
- The goal is to replace people rather than to stop wasting their evenings
- You need a strategy deck. We don’t make those, and we’d be bad at it.
The fastest way to find out is to talk.
Thirty minutes, no deck, no obligation. Bring the part of the week that eats you alive and we’ll tell you honestly whether it’s something we can fix.
Straight answer policy
If the honest read is that your problem is a process problem, or that you’re too early, we’ll say so on the call. A bad-fit engagement costs us more than it costs you.