How much does an AI agent cost in Monaco?
How much does an AI agent project cost for a company in Monaco?
The cost of an AI agent depends on the number of tools to connect, the state of your data, the volume handled, the business rules and the hosting. PRAGMA does not publish a price list: the audit and the working prototype delivered in 72 hours are free with no commitment, so you see a result working on your own data before any quote. The quote then covers what you saw working.
There is no single price list for an AI agent: two projects with the same name can involve very different amounts of work. This article explains what pushes a budget up or down, what an agent costs once it is running, and how to work out for yourself whether it will pay off, measured in hours saved rather than in promises.
Why PRAGMA does not publish a price list
We do not publish rates, because two agents with the same name can involve wildly different amounts of work. An agent that sorts email for a 3 person firm and several agents linked to the CRM, the accounting and WhatsApp of a 40 person company have almost nothing in common apart from the word agent.
- The quote comes after the audit and the prototype, never before.
- It covers what you saw working on your own data.
- It depends on the scope: tools to connect, volume, business rules, hosting.
- It itemises every line, so you know what you are paying for and why.
- The development phase takes 30 days, once the prototype is approved.
The AI transition, Productivity x Security, covers setting up agents connected to your tools, with access rights, human approval and a log of every action. The full offer is described on our AI transition in Monaco page.
What drives the price of an AI agent
An AI agent is a program that reads your information, applies your rules and prepares actions in your tools. Its cost depends mainly on what it has to reach and how precise it must be. These are the factors we price, roughly in the order in which they usually weigh most.
- The number of tools to connect: email, shared folders, CRM, accounting software, WhatsApp. Each connection has to be built, tested and secured.
- The state of your data: clean, structured files go quickly, while duplicated spreadsheets or scanned PDFs need clean-up work first.
- The volume handled: ten requests a day and several hundred do not call for the same architecture or the same monitoring.
- The number of rules and exceptions: sorting into three categories is simple, a workflow with special cases per client or per country is much less so.
- Access rights: when each person may only see their own files, those rights have to be modelled and checked.
- Hosting: European servers administered by PRAGMA today, with hosting in Monaco in preparation.
The factor that lowers a budget the most is still the choice of the first task. Starting with the one that costs the team the most hours each week, then moving on to the next, avoids paying for features nobody will use.
The running costs of an AI agent
The quote covers the build. Once the agent is live, three costs come back every month. We price them for your situation in the quote, because they depend directly on your volume and your hosting choices.
- Hosting: the servers that run the agent, store the action log and protect your data. The cost follows the amount of data and the level of availability you need.
- Language model usage: AI models are usually billed by the amount of text they read and produce. An agent that summarises three emails a day costs little, one that analyses hundreds of pages a day costs more.
- Maintenance: keeping up with updates to the connected tools, fixing a case that was handled badly, adjusting a rule when the way you work changes.
One point is worth checking with any provider: who owns the code, and what happens if you stop. At PRAGMA, the code belongs to you and a dedicated developer delivers every request tested.
What is free before any quote
Paying just to find out whether AI suits your business makes little sense. At PRAGMA, the first two steps are free and commit you to nothing.
- The audit: together we go through the team’s repetitive tasks, the tools it uses and the time each task takes every week.
- The 72 hour prototype: a working agent on your own data, which your team tries on real cases.
- The quote: it comes after the prototype and covers what you saw running, with a range explained line by line.
If the prototype does not convince you, you stop there. At the very least you leave with a costed list of the tasks that take up most of your time.
How to work out AI agent payback in hours saved
Payback is measured in hours given back to the team. The method takes five steps and fits in a spreadsheet.
- Measure the time spent on the task today, in hours per week, over two or three real weeks.
- Estimate the share the agent will take on, keeping the time a person needs to review and approve its work.
- Convert the quote into hours: its amount divided by the fully loaded hourly cost of the person concerned gives the number of working hours it represents.
- Convert the monthly running costs into hours the same way: hosting, model usage, maintenance.
- Divide the hours the project represents by the net hours saved each month, after running costs, to get the number of months before it pays for itself.
Hypothetical example, with figures chosen for illustration and unrelated to any client: an assistant spends 6 hours a week sorting email and preparing follow-ups. The agent takes over 4 of those hours, review included. 4 hours a week is about 17 hours a month (4 × 52 ÷ 12). If the quote represents 150 hours of that person’s work, it pays for itself in about 9 months (150 ÷ 17), a period the running costs lengthen.
Run the numbers with your own figures. The free prototype helps here too: it lets you measure the time actually saved on your cases instead of guessing. To choose which task to measure first, our article AI for small businesses: where to start gives practical criteria.
The controls included in the price
In every project, the agent only reaches the tools you open to it, a person approves before anything is sent, every action is logged, and your data is never used to train an external model. The rules that apply in Monaco, including Law No. 1.565, are covered in our article AI and personal data in Monaco.
What a first project with PRAGMA looks like
PRAGMA is a development studio based in Monaco, founded by Romain Girardi, an AI researcher who created EleutherIA, a system of AI agents, and Benjamin Mathias, tech lead with 8 years in infrastructure and DevOps, including at the European Space Agency. We built an AI-powered CRM for the Montblanc boutiques in Monaco and France, business software with AI agents for the insurance broker MediClar, and the AI back office of La Box Gourmande. Other projects are shown on our work page.
A first project always follows the same path: audit, 72 hour prototype, quote, 30 days of development, then go-live with a developer who stays with the project. To see what an agent can do in your line of business, our AI agents in Monaco page gives examples by sector.
Frequently asked questions
Why does PRAGMA not show prices?
Because the work varies too much from one project to the next for a price list to be honest. We would rather show you a free prototype on your data, then price exactly what you saw.
How do I know whether my project will be small or large?
The free audit tells you: it counts the tools to connect, the volume, the rules and the exceptions. A single agent linked to one tool stays a small project, several agents across several departments or offices make a larger one. The quote states where yours sits and why.
Is there a monthly fee on top of the project?
A live agent has recurring costs for hosting, AI model usage and maintenance. They depend on your volume and are priced in the quote, before you decide anything.
How long does it take for an AI agent to pay for itself?
It depends on the hours actually saved each week and on the team’s hourly cost. The method above lets you calculate it, and the free prototype lets you measure the saving on your own cases.
Is the 72 hour prototype really free?
Yes. It is built on your data, with no commitment. If the result does not suit you, you pay nothing and you keep the list of tasks identified during the audit.
Is an AI agent cheaper than hiring someone?
That is not the right comparison. An agent takes on repetitive tasks and a person approves its work. It frees time for the existing team to spend on clients and decisions.
For an estimate on your case, describe in the contact form the task that takes your team the most time and the tools it uses. We will come back to you for the free audit, then the 72 hour prototype, before any quote. You can also call +33 6 15 82 17 03.