![\"AI](\"https:\/\/storage.googleapis.com\/arrowlancer-prod.firebasestorage.app\/blog-ma\/prix-projet-ia-maroc-2026\/1.jpg\")
<\/p>\nAn AI project for a Moroccan SME costs, in 2026, anywhere from MAD 15,000 for a simple chatbot to more than MAD 300,000 for a custom assistant wired into your systems. That gap is exactly why most business owners postpone the decision: without a credible range, you cannot choose between a SaaS subscription, a dedicated build, or waiting.<\/p>\n
This article gives the prices actually charged on the Moroccan market, explains what pushes a quote up or down, and addresses the question few vendors raise on their own: are you allowed to have your customers’ data processed by an AI hosted abroad?<\/p>\n
You will leave with a reading grid and a checklist to scope a first pilot without buying custom work you do not need.<\/p>\n
Three things create the fog. First, “AI” covers very different things: an automated FAQ widget and an assistant that reads your contracts, queries your ERP and drafts replies share nothing in terms of effort. Second, part of the cost is recurring (API consumption, hosting, maintenance) and does not appear in the initial quote. Third, most vendors bill by the day without explaining how many of those days go into preparing your data, which is often the heaviest item.<\/p>\n
The result: a director receives a MAD 180,000 quote for something they pictured at MAD 30,000, or a MAD 400-per-month subscription that covers 20% of the use case. Both come from the same root cause: the need was not qualified before talking price.<\/p>\n
The figures below reflect what we see on SME and mid-size projects in Morocco. They are excluding VAT and include scoping, development and one production rollout, but not the recurring consumption detailed further down.<\/p>\n
<\/p>\n
A chatbot that answers frequent questions from your public pages and documents, with handover to a human, sits between MAD 15,000 and 40,000 when configured on an existing tool. As soon as it needs to check stock, an order status or create a ticket in your CRM, expect MAD 50,000 to 80,000. WhatsApp Business adds an integration plus per-conversation fees billed by Meta.<\/p>\n
Typical examples: extracting supplier invoice data into your accounting software, sorting and qualifying inbound emails, generating quotes from a form. Price depends mostly on how many systems must be connected and how variable the documents are. Invoice extraction across 3 supplier formats costs half of extraction across 40 formats.<\/p>\n
This is the most requested project in 2026: an assistant that answers staff questions from your procedures, contracts, product sheets or ticket history. The low end corresponds to a clean corpus of a few hundred documents; the high end to multiple sources (SharePoint, ERP, email), team-specific access rights and hosting in Morocco.<\/p>\n
Demand forecasting, fraud detection, image-based quality control, credit scoring: these need clean historical data, an experimentation phase and iterations. Below MAD 250,000, be wary: either the scope is very narrow or the data phase has been underestimated.<\/p>\n
Four items explain 80% of the difference between two quotes for a seemingly identical need.<\/p>\n
On top of these comes consumption: between MAD 500 and 10,000 per month in API calls for an SME, depending on the volume of conversations or documents processed. Always ask for an estimate on your real volume, not an average.<\/p>\n
A subscription to a general-purpose AI tool (office assistant, white-label chatbot, automation platform) costs MAD 200 to 3,000 per month, per company or per user. It is the right call when your use case is standard, your data is not sensitive, and you want to test whether teams actually adopt it before investing.<\/p>\n
Custom work is justified in three situations: your data is regulated or confidential (customer, health, HR, finance), the use case is core to how you differentiate, or the SaaS forces you to export data out of the country with no safeguards. We laid out this decision logic in our comparison of SaaS versus custom software in 2026<\/a>; it applies to AI with one nuance: custom AI does not mean training your own model, it means assembling existing building blocks around your data and your rules.<\/p>\n A middle path, often the most cost-effective for an SME, is to start with a SaaS tool on a non-sensitive scope for two or three months, then move only the process that proved its value onto a dedicated solution.<\/p>\n This is the point that, rightly, blocks many projects. As soon as an AI processes personal data (customer names, emails, phone numbers, HR files), you are within the scope of Law 09-08 and the CNDP. Two practical obligations follow: the processing must be declared or authorised depending on its nature, and transferring personal data to a foreign country is regulated and may require prior authorisation. The exact procedures evolve; check them on the CNDP website or with counsel before committing.<\/p>\n In practice, connecting a US-hosted AI assistant to your customer database with no contractual framework and no formalities amounts to organising a data transfer out of Morocco. It is not necessarily forbidden, but it must be framed. Our CNDP Law 09-08 compliance checklist<\/a> walks through the steps: data mapping, legal basis, informing individuals, a contract with the processor.<\/p>\n Three options exist, in order of increasing cost:<\/p>\n A good provider asks you this question in the first meeting, not after go-live.<\/p>\n A well-scoped pilot costs between MAD 25,000 and 60,000 and answers one question: does this work on our data, with our people? Here is the sequence we recommend.<\/p>\n This pilot fits naturally into a broader programme; if you are starting from scratch, our digital transformation roadmap for Moroccan SMEs<\/a> helps you place AI at the right moment rather than first.<\/p>\n The most common budget for a successful first project in a Moroccan SME lands between MAD 60,000 and 120,000 over the first year, pilot and maintenance included. Below that you are testing; above it you should already have a conclusive pilot behind you.<\/p>\n Arrowlancer builds AI solutions<\/a> for Moroccan companies: internal assistants, process automation, document analysis, with a sovereign option hosted in Morocco when your data requires it. If you have a process in mind and want an honest estimate rather than a pitch, describe your need to us<\/a> and we will tell you within a few days whether a pilot makes sense and at what price.<\/p>\n","protected":false},"excerpt":{"rendered":" Chatbot, internal assistant, automation: 2026 price ranges in MAD for an AI project in Morocco, what moves the quote, and the CNDP data-residency question.<\/p>\n","protected":false},"author":0,"featured_media":62,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[119],"tags":[100,103,106,63,109,112,115],"class_list":["post-61","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-cost_and_planning-en","tag-artificial-intelligence-en","tag-automation-en","tag-chatbot-en","tag-cndp-en","tag-morocco-en","tag-pricing-en","tag-sme-en"],"_links":{"self":[{"href":"https:\/\/arrowlancer.ma\/blog\/wp-json\/wp\/v2\/posts\/61","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/arrowlancer.ma\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/arrowlancer.ma\/blog\/wp-json\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/arrowlancer.ma\/blog\/wp-json\/wp\/v2\/comments?post=61"}],"version-history":[{"count":1,"href":"https:\/\/arrowlancer.ma\/blog\/wp-json\/wp\/v2\/posts\/61\/revisions"}],"predecessor-version":[{"id":64,"href":"https:\/\/arrowlancer.ma\/blog\/wp-json\/wp\/v2\/posts\/61\/revisions\/64"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/arrowlancer.ma\/blog\/wp-json\/wp\/v2\/media\/62"}],"wp:attachment":[{"href":"https:\/\/arrowlancer.ma\/blog\/wp-json\/wp\/v2\/media?parent=61"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/arrowlancer.ma\/blog\/wp-json\/wp\/v2\/categories?post=61"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/arrowlancer.ma\/blog\/wp-json\/wp\/v2\/tags?post=61"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}The CNDP question: where is your data when you plug AI into your processes?<\/h2>\n
\n
Billing traps and hidden costs to anticipate<\/h2>\n
\n
How to scope a first pilot in 4 weeks<\/h2>\n
\n
In short: budgeting an AI project in Morocco in 2026<\/h2>\n
\n
What now?<\/h2>\n