For an SME in 2026, getting into AI costs: an audit from €2,000 excl. VAT, a monthly AI Partnership at €399 or €699/mo excl. VAT with no commitment, and Custom Development from €3,000 excl. VAT. You can start small (an audit or a partnership) and ramp up from there — the right budget isn't the biggest one; it's the one that tackles first whatever pays for itself fastest.
An SME leader seriously weighing whether to make the move to AI quickly runs into the same wall: how much is this going to cost me? And then, the fog. Some talk about "a few thousand euros", others about six-figure budgets; half the websites display "contact us" where a price should be; and the online calculators demand your email before deigning to tell you anything at all.
This article puts the numbers on the table. Not a universal price grid — no such thing exists — but real ranges, line by line, from the first diagnostic to the return on investment, with ours displayed in plain sight. Enough to walk into any meeting, including one with us, knowing exactly what you're buying.
First reflex to correct, because it skews the whole reasoning: "getting into AI" is not a single expense you sign off on in one go. It's a sequence of decisions, each with its own entry ticket. You can perfectly well start with just one, measure its effect, then decide what comes next. The real budget risk isn't overpaying for a service — it's paying for the wrong one, at the wrong time, without knowing why.
How much does an AI Audit cost?
It's almost always the first expense — and the most profitable. A strategic AI Audit writes no code: it maps your company, identifies where AI creates value for you specifically, and puts a figure on the hidden costs you're absorbing without seeing them. In other words, it keeps you from spending in the wrong place.
At Codito Ergo Sum, the audit starts from €2,000 excl. VAT, in four packs depending on the depth you need (from the express diagnostic to the complete mapping of an organization with several business lines). What you get at the end: a map of your information flows, frictions quantified in euros and hours, a short-list of prioritized use cases and a roadmap. In practical terms, a document that turns "apparently we need to do AI" into "here are the three projects to launch, in this order, for this gain".
Why start there rather than with a tool? Because the most expensive line item in AI isn't development: it's the project you launch without knowing whether it serves any purpose. A €15,000 build on a process that creates no value costs infinitely more than a €2,000 audit that would have talked you out of it. The audit is the insurance against that particular waste — and it's also what quantifies the potential, and therefore justifies (or not) the spending that follows.
How much does a monthly AI Partnership cost?
The audit tells you what to do. The monthly AI Partnership makes sure it actually happens — month after month, without you having to carry it all on your own. It's the format that has won out for SMEs that want to embed lasting habits rather than stack up one-off engagements.
Two tiers, no commitment:
- Essential — €399/mo excl. VAT. A dedicated AI consultant, tailored training for the CEO, AI agents set up for their daily work, a monthly review and a direct channel. This is "the leader at the helm".
- Premium — €699/mo excl. VAT. Everything in Essential, plus a monthly training session for the teams and AI agents set up for employees. This is "transforming with the whole company".
The key point is the absence of commitment: you stop with a simple notice, the current month remains due, no extra month billed. You pay for as long as it serves you. That's what makes the expense legible in a budget: €399/mo is less than a professional software subscription, and you know exactly what to expect each month. Over a full year, Essential comes to less than €5,000 — compare that with the tens of thousands of euros a traditional consulting engagement costs over the same period.
For the full reasoning, we compared in depth the monthly AI Partnership versus the one-off consultant — where the one-off route typically runs around €6,000/month for a regular presence, onboarding and follow-up fees not included. The difference isn't just the sticker price: a partner who comes back every month compounds on your context, whereas a one-off consultant walks away with the knowledge they've built up.
How much does custom AI development cost?
When a use case is mature and deserves its own tool — an agent that handles your inbound requests, a business web app, an automation at the core of a process — you move to Custom Development. With us, it starts from €3,000 excl. VAT, quoted after a scoping session.
Why a quote rather than a catalogue price? Because four variables legitimately make the budget vary fivefold for an identical request: the complexity of the reasoning expected, the number of integrations with your information systems, the state of your data (structured, or ten years of scanned PDFs to clean up), and the level of interface (an agent inside a tool you already have costs far less than a full application with accounts and dashboards).
Concretely, the gap reads like this: a focused agent that plugs into a tool you already use sits at the low end of the range; a complete business application, with user accounts, permissions, dashboards and several connected systems, mechanically climbs. It's not "more or less AI" — it's more or less software around the AI. And custom doesn't mean starting from scratch: we assemble proven building blocks around your context, which keeps the cost contained.
One point too many quotes gloss over: an agent in production has recurring costs — model consumption billed by usage, hosting, ongoing maintenance (models change several times a year). For a focused agent with reasonable usage, expect a few dozen to a few hundred euros per month; more for a high-traffic platform. That's not a trap, it's a budget line: insist that it appears in the quote. We laid out the full pricing grid in our guide to custom AI development pricing.
Should you hire an in-house AI expert?
The temptation is real: what if we hired someone "to handle AI"? Let's run the numbers. A senior profile — Head of AI, Chief AI Officer, experienced AI developer — commands around €80,000 gross/year in France, which, payroll charges included, means well over €100,000 for the company. Before touching a single project.
On top of that cost come the blind spots of recruiting: a position that's hard to source, hard to assess when you're not an expert yourself, and risky (a bad hire on this profile means six months and a budget down the drain). For an SME, hiring full-time for a skill whose need hasn't yet stabilized amounts to paying for wildly overpriced insurance.
The alternative is shared-time expertise. A partnership at €399 or €699/mo comes to between €5,000 and €8,500 over a year — the order of magnitude of a single month of a Head of AI's fully loaded salary, for continuous access to the same expertise, without the hiring risk. In-house recruitment becomes relevant later, once the volume of projects clearly justifies it. Outsourcing first is precisely what tells you if and when that moment arrives.
What AI budget for an SME just getting started?
No need to add everything up. The healthiest trajectory, for an SME starting from zero, looks like this:
- Month 0 — an audit (from €2,000 excl. VAT). To know where to aim before spending. It's the investment that makes all the others pay off.
- Month 1 and onward — a partnership (€399 or €699/mo excl. VAT). To execute the roadmap, train people, set up the first agents, with no commitment.
- When a use case is mature — a focused build (from €3,000 excl. VAT). At the right moment, on the right process, once the value is proven.
An SME can therefore set a genuine AI dynamic in motion for a few thousand euros in the first year, then invest more once the first gains are visible. It's the opposite of the pharaonic project launched blind: start small, prove it, reinvest what you've gained.
What ROI should you expect?
The only question that really matters. An AI budget isn't a prestige expense: it has to pay for itself. And when the scope is well chosen, it does so quickly.
On a concrete client case — Amourdedieu, through its subsidiary VLT — the audit identified €636,000 in hidden costs that were weighing on the company without ever having been quantified. Once the first AI projects were launched on the right areas, the return on investment materialized in about 2 months. The cost of the audit and the first actions was covered almost immediately by what they were saving.
The lesson isn't "AI always returns €636,000" — every company is different, and one case doesn't make a guarantee. It's that the return comes from prioritization: tackling the most expensive frictions first, rather than sprinkling AI a little bit everywhere. A euro invested in the right place pays off; the same euro spent on an impressive gadget returns nothing. That's exactly what a serious audit brings to light, and what separates an AI budget that pays for itself from one you merely endure.
And the return isn't only about money saved. It's also measured in time freed up for the teams, in mistakes avoided, and in the ability to handle more volume without hiring. Those gains are more diffuse, but they compound month after month — one more reason to embed the habits for the long run rather than aim for a one-off win.
In summary
AI for an SME in 2026 is not a six-figure blank check. It's an audit from €2,000, a partnership at €399 or €699/mo with no commitment, a build from €3,000 when the time is right — and far less risk than an €80,000-a-year hire. The right budget isn't the smallest or the biggest: it's the one that starts by knowing where to aim.
If you'd rather put numbers on your situation than on averages, that's precisely what the first meeting is for.