SME Automation in Luxembourg: Fund It and Ship It in 90 Days
For: Luxembourg SME owners funding and rolling out a first automation in about 90 days
For: Luxembourg SME owners funding and rolling out a first automation in about 90 days
The whole idea in one line
A Luxembourg SME can have the State reimburse 70% of a first automation through SME Package - AI, then sequence the whole build around the aid decision. Get the funding mechanics and the calendar right, and your first automation goes from quote to live system to reimbursement in roughly 90 days, without spending out of order.
In short: it means having the State reimburse 70% of an eligible first automation project through SME Package - AI, on a project worth EUR 3,000 to 25,000 excl. VAT, paid after the work is done (Guichet.lu). It is real money, it is not automatic, and the part nobody explains is how to sequence the build around the aid decision.
Here is the honest problem. Search “SME automation Luxembourg” today and you get two kinds of page. The first is an agency that tells you the SME Package - AI exists, dangles the EUR 17,500 number, then hides the mechanics behind an eligibility quiz so you call them. The second is a generic “how SMEs can use AI” guide with no Luxembourg funding detail at all. Neither answers the question the owner is actually sitting with: concretely, how do I get the State to pay most of this, and what does the calendar from application to live system to money-back really look like?
“The funnels want you to call them and the explainers do not know the local rules. The thing in between, the plain funding mechanics plus the order you do things in, is what almost nobody writes down.”
This article is the in-between. It covers the funding mechanics and the funding-gated rollout, and it deliberately stops there. It does not tell you which process to automate (that is a different decision, covered in our guide to choosing which process to automate first), it does not do the ROI maths, and it does not argue rules-versus-AI. If the workflow is lead handling, define how the lead gets qualified before you fund software around it. It assumes you have chosen the workflow and decided it is worth doing, and answers the one question those leave open: how do you fund it and ship it.
Short answer: for most first automations, SME Package - AI is the entry track. It is built for one small, well-defined AI project in the EUR 3,000 to 25,000 band (Guichet.lu). Broader digitalisation efforts route through Luxinnovation’s Fit 4 Digital and Fit 4 AI programmes (Luxinnovation). Pick by scope: one workflow versus a wider programme.
The mistake I see owners make is reaching for the bigger programme because it sounds more serious. For a first automation it is usually the wrong reach. SME Package - AI exists precisely so a small firm can fund one contained tool without committing to a transformation. Start where the bar is low and the project is real.
The entry track for a small, well-defined AI project. The completed project must be worth between EUR 3,000 and EUR 25,000 excl. VAT, and the Ministry of the Economy reimburses 70% of eligible costs after the package is in place (Guichet.lu). You start with a pre-analysis at the House of Entrepreneurship or the eHandwierk department.
Luxinnovation-run pathways for SMEs that want guided support across a wider digitalisation or AI effort, not just one tool. Useful once the first automation has proven itself and the next step is a programme rather than a single workflow. Eligibility is checked with Luxinnovation directly (Luxinnovation).
Both tracks share the same logic: the State helps because you are doing something specific and useful, not because you filled in a form. The funding is generous (Guichet.lu), and it is granted as de minimis aid under EU Regulation 2023/2831, which is worth knowing if your firm has already received other aid this cycle. For the rest of this guide I assume the SME Package - AI route, because that is the one a first automation almost always belongs in.
Short answer: you qualify if you hold a business permit from the Ministry of the Economy, meet the SME criteria on staff, turnover and balance sheet, and have your registered office in Luxembourg (Guichet.lu). Before you apply, a free pre-analysis with the House of Entrepreneurship checks your AI readiness and shapes the priority actions your file is built on.
| Eligibility test | What it means in practice |
|---|---|
| Business permit | A valid permit issued by the Ministry of the Economy. |
| SME criteria | You meet the staff, annual turnover and annual balance-sheet conditions for an SME. |
| Registered in Luxembourg | Your registered office is in Luxembourg. |
| A real project | A defined AI project worth EUR 3,000 to 25,000 excl. VAT, not a vague intention. |
The pre-analysis is the step owners underrate. It is run through the House of Entrepreneurship, or the eHandwierk department of the Chamber of Skilled Trades and Crafts for craft businesses, and it looks at where your firm actually is with AI and which actions matter first (Guichet.lu). Treat it as the moment your funding file gets its spine, not as a box to tick. A free maturity read of this kind is also offered through the Luxembourg Digital Innovation Hub ecosystem — the practical point is that you do not pay to find out whether you are ready, and the output strengthens the application.
I will give you the real engagement this draws from, anonymised, because the lesson is exactly the readiness point. I built a custom AI content-operations assistant for an independent company director I worked with — I will not name the person or the tool. We did not start with a suite. We started with one bounded, repetitive job: idea capture. The system pulled from websites, videos, PDFs, an RSS feed and a curated watchlist, and the AI extracted content ideas from the lot. That one workflow had to work before we touched anything else. Only once it did, did we expand it into production (scripts, slides, formatted documents, transcripts) and online presence (a posting scheduler, a knowledge base). That sequencing, one painful workflow proven first, is what a strong maturity read points you toward, and it is the same shape a first SME Package - AI project should take.
Short answer: the order is pre-analysis, then define the package and pick a provider, then an adviser files the application, then the Ministry of the Economy reviews and decides, then implementation, then evaluation and the 70% reimbursement (Guichet.lu). Before you pick that provider, the guide to choosing a business advisor for a Luxembourg SME is the useful companion. The review is the wait you must plan around; the build only starts once the decision lands.
Pre-analysis
You contact the House of Entrepreneurship (Chamber of Commerce) or the eHandwierk department (Chamber of Skilled Trades and Crafts, for craft firms). They run a pre-analysis of your current AI situation and identify priority actions. This is also where the free maturity view comes from.
Define the package and pick a provider
You choose the package and the provider that fits, and together agree a quote between EUR 3,000 and EUR 25,000 excl. VAT. The scope you write here is the scope the State assesses, so keep it narrow and real.
Submit the application
A House of Entrepreneurship or eHandwierk adviser completes and files the financial-aid application for you. You do not file alone.
Review and decision
The Ministry of the Economy analyses the file and decides. This is the wait you cannot rush, and the part owners most often forget to plan around.
Implementation
Once you have the green light, the provider implements the AI solution. This is the build, and it is the only phase where most of the spend lands.
Evaluation and reimbursement
An evaluation meeting reviews satisfaction and savings, the process is closed, you pay the final bill, and the Ministry reimburses 70% of eligible costs under the SME Package - AI rules (Guichet.lu).
The one tax this market really charges is patience, and the review step is where you pay it. The official pages set out the procedure but not a guaranteed turnaround, so plan as if the decision takes weeks, not days, and build your calendar around it rather than assuming you can start the day after you submit. That single assumption, that the build waits for the decision, is what the next section turns into a working calendar.
Short answer: sequence the build around the aid lifecycle. Prepare the file and the quote before approval, hold paid build work until the decision lands, deliver in a focused window once approved, then go live and trigger the 70% reimbursement. The point is what you can do early versus what must wait, so a first automation ships in about 90 days.
Book the pre-analysis. Write the one workflow you are funding in one paragraph. Get the provider quote inside the EUR 3,000-25,000 SME Package - AI band (Guichet.lu). You can do all of this before any decision, and you should, because a tight file moves faster.
The adviser files the application; the Ministry reviews. Do not start paid build work here. If you spend before the decision, you put the eligibility of those costs at risk. Use the wait to prepare data and brief the team, not to start the cheque-writing work.
With the decision in hand, the provider implements the solution on the single workflow. Keep the scope identical to what was approved. This is where the work actually happens, and a contained first automation lands well inside a month of build.
The solution goes live, the evaluation meeting confirms satisfaction and savings, you pay the final bill, and you trigger the 70% reimbursement. The aid arrives after, so plan the cash to front the full outlay first.
Read the calendar as one rule with four faces: money comes back after, so cash goes out first, and nothing you pay for happens before the decision. The reimbursement of 70% is triggered only once the package is in place and the bill is paid (Guichet.lu). That is why the prepare phase is free work you should front-load, the wait phase is for everything that costs nothing, and the build phase is short because the scope was kept to one workflow.
My job in your rollout is not to be the hero of it. It is to be the guide, and the gift I bring is this sequencing. With the director’s content-ops assistant, the thing that made it work was never the technology — it was building one workflow, proving it, and only then expanding. A funded first automation is the same discipline with a State decision sitting in the middle of it: prepare, wait without spending, build narrow, then claim.
Short answer: the common failures are scoping too broad, starting paid build work before the decision lands, treating the pre-analysis as a formality, and budgeting only for the net 30% when you front the full bill. None of these is about the technology. All of them are about respecting the order of the SME Package - AI aid.
A first application that tries to automate three departments at once is slower to approve and harder to deliver. One workflow, one quote, one clear outcome. Breadth is for the second project, not the first.
The reimbursement is paid after the package is in place, but the costs still have to be eligible. Starting paid build work before approval is the fastest way to put your own aid at risk.
The pre-analysis is not a hoop. It shapes priority actions and strengthens the file. Walk in with a real workflow, not a vague wish to "do AI", and the rest of the process is calmer.
You pay the full bill first and recover 70% under SME Package - AI later. An owner who budgets only for the net 30% gets a cash surprise. Size the project against the full up-front outlay.
“A first automation does not fail on the model. It fails when an owner spends out of order, scopes too wide, or budgets for the 70% they get back instead of the 100% they pay first.”
Short answer: once the first project is live and reimbursed, the next move is a deliberate one, not a reflex. Choose the next process by impact, build the ROI case before you commit, and decide whether the job needs plain rules or genuine AI. Each of those is its own decision, and each has a guide.
Pick by where the pain is heaviest and most repetitive, not by what is fashionable.
How to choose which process to automate firstScore the payback before you spend, especially since you front the cash and recover later.
Automation ROI for Luxembourg SMEsSome jobs need a rule, not a model. Knowing which keeps the next project honest.
What AI adds over rules-based automationThat is the payoff of funding the first one well. You have a live system, most of the cost back, and a team that watched you sequence a real project responsibly. That standing is what lets you fund a second, bigger automation — and if you want help sizing how much to risk on it, our guide to the affordable-loss test for a first AI budget applies the same survivable-first logic to the spend.
Stepping back, this is also the cleanest test of when AI support is worth the effort: when one workflow has been proven, the team has learned the discipline, and the next investment has a measurable baseline to beat rather than a vague aspiration to chase.