Custom micro SaaS · Finance & ERP · AI · People systems
We learn how your business runs, then design, build and run the technology it needs: finance and ERP, operations, people tools and AI. One partner, end to end.
Built for your business
Every business runs a little differently: how orders arrive, how stock moves, who approves what, how shifts get filled, when invoices go out. Generic software makes you carry that difference in spreadsheets and workarounds. We build around it instead.
End-to-end solutions
Agencies take a brief and write code. We start with how money, stock, people and information actually move through your business, curate the solution it needs, then provide the technology and keep it running. One partner from problem to working system.
We learn the operation from the people running it: margins, bottlenecks, seasonality, who decides what and where the hours go. The picture comes before any technology.
Build, buy, integrate or simplify. We recommend what fits: custom software where it pays, proven tools where they are good enough, and a process fix when code is not the answer.
Then we deliver it end to end: build, integrate, host, train the team and improve it as the business changes. One accountable partner and one number to call.
Services
Operations, finance and people: the unglamorous systems that decide whether a business runs smoothly. We build them around how you work, connect them to what you already use, and keep them running. Each one shaped to your business, never a template.
Purpose-built software we design, build, host and run for one business: yours. No licence sprawl, no half-fit templates, one team accountable for the whole thing from the first workshop to the thousandth release.
Invoice capture, three-way matching, bank reconciliation, collections and a month-end close that runs on rails, posted cleanly into the ledger you already keep.
Right-sized ERP for companies that have outgrown spreadsheets but not their budget: custom modules, clean migrations, and integrations with the ERP, accounting and payroll systems you keep.
Demand forecasts, reorder points, supplier lead times and order flows across every channel, built on your real sales rather than last year's spreadsheet.
Quoting, B2B order portals, D2C storefront operations, returns and channel sync, with the CRM follow-ups that keep the pipeline moving.
Onboarding, rosters, attendance and leave, timesheets, payroll inputs and reviews, for teams on the floor, in the field or across several sites.
Agents that read documents, chase approvals and answer questions on your data, with a person in the loop wherever money moves.
Forecasting, segmentation, anomaly detection and scoring models tuned to small-business data volumes, and explained in plain language.
Dashboards and digests operators actually open: health scores, margin by product and customer, cash conversion and what needs a decision today.
Approvals, handoffs and notifications wired across the tools you already use, so work moves without someone pushing it.
Solutions
A few of the shapes our work takes. Each one usually starts life as a spreadsheet, a group chat, or a person who knows too much.
01 · Supply chain
Demand forecasts per SKU and channel, reorder points that respect supplier lead times and minimum order quantities, and a Monday-morning list of exactly what to order.
| SKU | Cover | Lead time | Action |
|---|---|---|---|
| 4471 · Cotton tote | 6 d | 14 d | Order 480 |
| 2210 · Glass bottle 750 | 3 d | 21 d | Expedite |
| 3390 · Gift box M | 11 wk | 10 d | Hold |
02 · Analytics
One score that rolls up margin, cash, receivables and inventory, with the drivers underneath it. Refreshed nightly from your accounting, commerce and bank feeds, delivered as a Monday digest.
03 · Finance automation
Invoices in and out, payments and receipts, bank lines and the ledger behind them, kept in step automatically. Documents are read and coded, payments matched to invoices, bank feeds reconciled every night, and only the exceptions reach a person.
| Document | Party | Amount | Status |
|---|---|---|---|
| BILL-2041 | Meridian Packaging | 12,480.00 | Paid |
| INV-0917 | Harbor Retail | 8,900.00 | Overdue 12 d |
| BILL-2043 | Kalon Textiles | 3,215.50 | Tax check |
| INV-0918 | Northline Stores | 1,120.00 | Paid |
06:00Bank feed: 214 of 216 lines reconciled
09:14INV-0917 reminder sent · due 12 days ago
09:15BILL-2043 tax rate differs from order · review
10:02Payment run: 9 bills scheduled for Friday
04 · Process automation & agents
Orders from every channel land in one flow: risk and address checks, stock allocation, the 3PL handoff, customer updates and ledger entries, with an agent taking the judgement calls a rule cannot.
#48213Order 10492 · dispatched via 3PL · 2m 14s
#48212Order 10491 · dispatched · customer notified
#48211Order 10490 · address ambiguous · agent asked the customer
#48210Order 10489 · split shipment · ledger updated
05 · People management
One place for shifts, leave, attendance and the numbers payroll needs, built for teams on the floor, in the field or across several sites. Managers approve from their phone; payroll gets a clean file instead of a chase.
| Shift | Site | Staffed | Status |
|---|---|---|---|
| Sat · late | Warehouse B | 5 of 6 | Needs 1 |
| Sun · early | Store 2 | 4 of 4 | Filled |
| Mon · day | Field crew A | 6 of 6 | Filled |
| Mon · day | Head office | 11 of 12 | Leave covered |
AI · Machine learning · Deep learning
A regression when it works, a fine-tuned small model when it pays, a frontier language model when the job needs judgement. We pick the technique for the problem, prove it on your own ERP, accounting, commerce and HR data, and keep it honest in production.
Demand, cash flow, staffing and lead-time forecasts with honest uncertainty bands, hierarchical across SKU, channel and region.
Reorder prompts, cross-sell and next-best-action suggestions for sales and service teams, learned from your own transactions and behaviour rather than a generic catalogue.
Assistants grounded in your documents and databases, and tool-using agents that carry a workflow end to end, with guardrails, evaluations and a person on the loop.
Margin leaks, duplicate payments, unusual orders and slipping customers caught early, scored and explained so the team knows what to do next.
Stock counts from shelf photos, packaging and label checks, and quality inspection on a production line, trained on your images and run on modest hardware.
Customer segments, product affinities and price and promotion elasticity that turn a sales history into decisions about range, bundles and discounts.
Drift monitoring, scheduled retraining, evaluation sets and rollback. Small-team MLOps that runs itself.
Every forecast, score and extraction carries the reasons behind it, written for operators rather than data scientists.
Your data is never used to train shared models. Private deployments and region-appropriate hosting where compliance needs them.
Methodology
Every engagement follows the same discipline: map how the business actually works, find where it hurts, scope a solution with the expected result written down, then track the numbers that show whether it worked and act when they move.
We map how orders, money, people and information move through the business, learned from the people doing the work. Systems, spreadsheets and workarounds included.
OutputA system map everyone recognises
Where hours leak, where data breaks, where decisions wait. Ranked by what each one costs the business, not by who mentions it loudest.
OutputA ranked list with a cost against each item
Each fix comes with the result we expect, in numbers, and the smallest build that gets there. You see the expected return before anything is built.
OutputA scoped plan with targets and a price
Once live, the metrics we agreed are tracked in the open, week by week. When a number moves the wrong way we change the software or the process, not the report.
OutputLive metrics and a response, not a slide deck
What the metrics show feeds the next round of understanding, so the loop never closes on a guess. Targets are written into every scope. For example:
How an engagement runs
We sit with the people doing the work. We map where hours leak, where data breaks and what fixing it is worth, then write it down on a page you can argue with.
A thin, opinionated spec and a clickable prototype. You see the screens and the integrations before anyone writes production code.
Weekly releases into your hands, integrations first and polish second. Nothing you did not ask for, nothing you will not use.
We host, monitor, patch and improve. You get a person to call, a monthly number, and a roadmap that follows the business rather than the other way round.
Why NaOHX
NaOHX is an end-to-end solutions partner for small businesses with serious operations, not a software agency. Engineers, data scientists and operators who understand the business first, curate what it actually needs, and then build and run the technology behind it.
Talk to the teamScoped and priced for companies with 5 to 200 people. No enterprise theatre, no six-month discovery phases, no seat licences that punish you for growing.
We learn the process from the people running it before we write a line. The software fits the work, not the other way around.
Code, data, infrastructure accounts. Exportable on request and documented as we go. We earn the next month by being useful, never by locking you in.
Audit trails, permissions, backups and monitoring come standard. AI does the reading and the drafting; a person signs off wherever money moves.
FAQ
Straight answers. If yours is not here, ask it on a call.
A small, purpose-built piece of software that does one job for your business very well, which we design, build, host and keep running for a monthly fee. You get the fit of custom software with the predictability of a subscription.
Agencies take a brief, hand over code and leave. Off-the-shelf tools make you bend your process to theirs. We start with the business and recommend what it actually needs, whether that is software we build, tools you already own connected properly, or a process fix with no code at all. Then we stay accountable for the technology in production, shaped around how you already work.
Yes. Purpose-built does not have to mean expensive: we keep each piece small, reuse the boring infrastructure across clients and spend the budget on the part that is specific to you. The result is software that fits your business, not a generic product you have to fit around.
Two ways. Fixed-scope sprints for well-defined problems, quoted in writing after the diagnosis. Or a monthly build-and-run retainer for products that keep evolving. Either way the number is agreed before work starts.
You do. Everything runs in accounts you control or can take over, the code is yours, and your data is exportable on request. We earn the next month by being useful, not by locking you in.
That is usually the point. Commerce platforms, marketplaces, accounting software, warehouse and logistics systems, spreadsheets and messaging apps: we integrate first and replace only when it pays.
Rarely. Most small businesses need the layer around those systems: approvals, reconciliation, forecasting, rostering and HR tools, the dashboards. We build that layer and connect it to the ERP, accounting and payroll software you keep. When a system genuinely has to go, we plan the migration in stages so nothing stops.
AI does the reading, drafting and classifying. A person approves anything that moves money or reaches a customer. Every automated decision is logged with its evidence, your data is never used to train shared models, and hosting follows your region and compliance needs.
Typically founder-led or family-run companies with 5 to 200 people and a real operation to run: stock, orders, invoices, suppliers. If you are earlier than that, we will tell you what to buy off the shelf instead.
Contact
Thirty minutes is enough to tell whether we can help. If we cannot, we will say so and point you somewhere useful.
