WhatsApp Business
Bookings, reminders, order updates and support routing on WhatsApp's own business platform — with a clean handover to a person the moment the conversation needs one.
Custom software and automation for companies of ten to two hundred and fifty people, wherever they are. Fixed scope, fixed price, handed over with the source code and the documentation.
Twelve sites. Guests confirm, move or cancel in the thread they already use. Measured over the first quarter.
No accounts to create, no card details stored anywhere, nothing for the customer to install.
Not a menu of everything. These are the jobs that recur in companies of this size, and the ones worth automating first.
Bookings, reminders, order updates and support routing on WhatsApp's own business platform — with a clean handover to a person the moment the conversation needs one.
Two hundred CVs, a month of invoices, an inbox that never empties — read, sorted and filed while you do something else. A person signs off before anything is committed.
One screen, with permissions, that replaces the spreadsheet four people edit at once.
Booking, quoting, onboarding, document exchange. Hosted, monitored, and yours.
The recurring manual job — the export, the reconciliation, the Monday report — running on its own and telling you when it doesn't.
Each entry says where it stands. Delivered means finished and running for a real client. Not yet delivered means the same kind of job, not yet commissioned by anyone.
Guests confirm, move or cancel in the thread they already use. No-shows fell by roughly a third over the first quarter. Built on expressai.bot, the WhatsApp assistant platform behind it.
Customers pick from the day's menu, build an order and send it straight to WhatsApp — no accounts, no payment gateway, no card details stored anywhere. miumarmitas.com
CV intake for recruitment · production and stock reporting · shift scheduling with an automatic payroll export.
Client names are deliberately omitted throughout.
Rooms, elements, phases and photography in one place, replacing three spreadsheets that had to be kept in step by hand. Progress is stated as delivered sub-projects against the total, so the figure can be checked rather than taken on trust. Full project breakdown
What you want removed, roughly how often it runs, and who touches it. That is enough to answer.
Within two working days: a scope with a fixed price, or a straight explanation of why it isn't a fit and what would be.
From a few days for a single automation to twelve weeks for a full build. Handed over with source code and documentation you own.
Three fixed packages, and one for work that has to be scoped before it can be priced. Changes after delivery are billed by the hour at a rate you know in advance — no surprise invoices.
One automation, end to end.
A working first version, or automation connected to the systems you already run.
Several systems or sites tied together, with a roadmap rather than a single delivery.
A sector with rules about where data is allowed to live, or a scope too particular to put a number on before we have talked.
Because the work takes fewer hours than it used to. AI handles the repetitive part — the groundwork, the first drafts, the routine checks — so a project that once needed a team of four for three months can be done properly by one senior engineer in weeks. The saving is in hours, not in standards. Architecture is still decided before any code is written, security is still reviewed, the work is still left clean enough for someone else to pick up, and it is still checked before it ships. These prices sit under what an agency would quote for the same scope — by the margin the hours explain, not by a margin that should worry you.
AI is a tool in the process, the way a power tool is in a workshop. It is supervised and orchestrated by a software engineer who reads every line, decides the structure, and is answerable for what ships. Nothing reaches production because a model produced it. If that distinction did not hold, the price would be the only thing you were buying.
Yes, and that is what most people should do. Whatever basic delivered is credited in full — you never pay twice for the same code. Upgrade within three months and you pay the difference between the two tiers. After that it is quoted, because building in two phases takes longer than building once: a second round of scoping, deployment and testing. If you are on cloud, the monthly moves to the rate for the new tier from the month the upgrade goes live.
The source code, the documentation, and the right to do whatever you like with both. No licence, no per-seat fee, no permission needed to hire someone else to change it. If you host it yourself, there is nothing left to pay.
Keep it in-house if your data cannot leave the building. It does not need a server room — it can run on a computer or laptop you already have. Let us host it if you would rather not think about it: it costs more because it includes the hosting, the backups, the updates and someone keeping an eye on it. Most companies of this size have us host it.
You own the source code and the documentation from day one, and both are written so another developer can pick them up without an introduction. Marta Oryl holds the relationship, not an individual, and a second engineer is already familiar with the work. Nothing here depends on one person staying reachable forever.
Roughly how often it runs, and who touches it. That is enough to answer. You get a reply within two working days: either a scope with a fixed price, or a straight explanation of why it isn't a fit and what would be.