A fourth way for SMEs
For years, an SME manager who wanted a tool suited to the business had three options. Take a generic piece of software and bend the organisation to its screens. Order custom development, long, expensive and hard to evolve. Or improvise a workaround: a shared spreadsheet, an email chain, a notebook at the counter.
Each of these options has a hidden cost. Generic software imposes its way of working. Custom development ties up time and money before the first result. The manual workaround relies on one person who knows, and collapses when that person is away.
A fourth way now exists. You describe the process that slows your company down, in your own words, and AI assembles a working application on a foundation that is already governed. It is not a magic promise. It is a method, with steps, checks and people who validate. This article describes it as it actually happens.
Between your description and the application
The most useful question is not “can AI do it?” but “what exactly happens between my description and the application in production?”. Here is the journey.
- You describe the expected result
Not the technology, the result: what must be tracked, who uses it, in what setting, under which constraints. A technician on site does not have the same needs as an assistant at the office.
- Proven business blocks are assembled
The application is not written from scratch. It is composed from objects, screens, rules and roles already in use elsewhere. AI selects, adapts and connects those blocks to your description.
- Normal cases and exceptions are tested
The everyday case, the edge cases, the permissions of each role and the ability to roll back are checked before anyone opens the application.
- Your team checks it in a validation environment
The people who will use the application test it with real situations. They flag what is missing and what gets in the way. Adjustments are made at this stage.
- Nothing goes to production without your approval
Going live is a decision taken by your company, not an automatic event. You approve, then the application is opened to its users.
- The application is operated and enriched
After going live, real use reveals what needs to evolve. The next requests follow the same journey: description, assembly, test, validation.
This journey has a simple consequence: at no point do you receive a tool that nobody in your company has seen working.
What AI does and what your experts keep
The question comes up in almost every conversation with a manager: does AI replace developers? The calm answer is this. AI lowers the cost of standard tasks. Data entry screens, lists, status tracking, passing information from one person to the next, notifications: all of that becomes fast and inexpensive to assemble.
What stays in human hands has not changed. The context of your business, which nobody else knows. The responsibility for what the application does to customers and accounts. The relationship with the people who use it. And the decisions that commit the company: what is approved, by whom, above which threshold.
There is another difference with what is often called a prototype. An application assembled this way is not an isolated demonstration on a laptop. It lives in the same system as your customers, your invoices and your tickets. It reads the same data, applies the same access rights, leaves the same record. It is a durable capability, not a trial.
AI assembles what is standard. Your teams keep what makes your company.
Choosing the first project
The first project decides a lot. If it is too broad, it drags on. If it is too abstract, nobody can say whether it succeeded. Good first projects look alike: one process that slows you down, few people involved, a clear result and a manual workaround that already exists.
The manual workaround is an excellent sign. If a shared spreadsheet or an email chain does the job today, the need is real, the vocabulary is known and the rules are already applied, even if fragile. A few frequent examples:
- Intervention requests: a customer reports a problem, someone assigns it, a technician steps in, the customer is kept informed.
- Quote approvals: above a certain amount, a second person must approve before sending.
- The onboarding checklist for a new customer: documents to receive, access to create, people to notify, without forgetting anything.
- The equipment register: what is installed where, since when, under which contract and with which next deadline.
Choose the one that costs the most time to your most solicited person. That is where the result will be visible soonest.
What to prepare before you start
You do not have to write anything technical. However, a few elements make the description much more precise, and the application much more accurate from the first version.
- The vocabulary of your business: what you call a request, a file, an intervention, a status. The application must speak your language, not the other way round.
- An example of a real document: a quote, an intervention sheet, a typical email. One example is worth ten explanations.
- The rules and the exceptions: what happens normally, and what happens when the customer is late, when the technician is away, when the amount exceeds a threshold.
- Who validates what: the person who approves, the one who can edit, the one who only consults.
We detailed how to phrase that description, with examples, in Describing your need to AI without being a technician.
Neoo Forge: the workshop that assembles
At Neoo, that workshop is called Neoo Forge, the workshop that assembles your applications with AI. It is part of the Neoo Business OS, the system that runs your whole company: sales, finance, operations and support in one place.
That is what changes everything for the application you describe. It is assembled on the same foundation as your customers, your invoices and your tickets. So it is connected from day one: an intervention request knows the customer concerned, a quote approval sees the real amount, the equipment register finds the associated contract. No double entry, no export to redo.
Every change follows the same validation journey, and every version can be found again. We explain that mechanism in Change, validate, roll back. And if you want to understand why this shared foundation comes before everything else, What is a Business OS sets the frame.
To get started, two paths. Describe your idea on How much would my idea cost? and get a first estimate. Or discover the workshop in detail on the Neoo Forge page.
Frequently asked questions
Do I need technical knowledge?
No. You describe the process in your own words, as you would to a colleague. What matters is your knowledge of the business: the vocabulary, the rules, the exceptions and the people who validate. The technical assembly is done by AI and verified through test steps.
Can the application be changed after it is built?
Yes, and that is planned from the start. Real use always reveals adjustments. Every change request follows the same journey: description, assembly, test, validation by your team. Previous versions remain available to roll back if needed.
Does it work with the tools we already use?
Connections with your existing tools are defined and validated with each customer before deployment. We look together at what must be connected, in which direction and under which rules, rather than promising general compatibility.