Building X++ without a consultant: myth or reality in 2026?

X++ is the proprietary programming language of Microsoft Dynamics 365 Finance & Operations (formerly AX). It is an object-oriented language derived from C++, but with peculiarities that make it nearly incomprehensible to a .NET or Java developer without specific training. Base classes, tables as first-class objects, the MorphX framework, extension chains (CoC — Chain of Command)... all concepts that cannot be learned in a week.

This complexity has long justified high rates for X++ developers. In the French market, there are fewer than 2,000 experienced X++ developers, compared to several hundred thousand JavaScript or Python developers. Scarcity creates cost, and D365 integrators have known how to leverage this.

Add to this the development environment: Visual Studio integrated with LCS (Lifecycle Services), cloud deployment models, compilation cycles that can take several minutes, and sandbox environments to manage... The learning curve is real and is unlikely to disappear for complex developments.

Large language models (LLMs) have been trained on significant volumes of X++ code: Microsoft documentation, GitHub repositories, technical forums, and case studies published by partners. For common patterns, they are now remarkably effective.

A well-prompted LLM can reliably generate: standard class extensions (adding fields, overriding methods via CoC), simple to medium batch jobs, basic SSRS reports, Data Entity-based integrations, custom workflows following standard Microsoft patterns, and data migration scripts. For these use cases — which represent approximately 70% of common development tickets — AI can produce functional code in minutes.

The key is in the prompting and validation. A D365 expert who knows how to precisely formulate the need, provide the context of the existing data model, and review the generated code can deliver these developments in a fraction of the traditional time. This is exactly SKALP AI's model: AI generates, the expert validates, the CIO approves in Azure DevOps.

It would be dishonest to claim that AI can do everything. Complex X++ developments — real-time integrations with critical legacy systems, deep customizations of the workflow engine, performance optimization on databases of tens of millions of rows — still require human expertise that AI alone cannot guarantee.

Similarly, situations where the problem is poorly defined — where the business does not yet know exactly what it wants — require a consultant capable of running requirements-gathering workshops, challenging assumptions, and proposing architectural alternatives. AI excels at executing a precise requirement; it does not replace an expert's ability to define that requirement.

Finally, heavily customized D365 environments, with years of undocumented customizations piled on top of each other, represent a particular challenge. AI needs context to generate relevant code. Without documentation, without a clear data model, the quality of generation degrades. This is why governance — documentation, Azure DevOps, code reviews — remains a prerequisite, not an option.

The real question is not 'does AI replace the consultant?' but 'how do we combine AI and human expertise to divide D365 development costs by 10?'. SKALP AI's answer is clear: AI handles generation (code, specifications, unit tests), the D365 expert handles technical validation and quality prompting, and the CIO retains full control via Azure DevOps.

This model does not require your internal teams to learn X++. Nor does it require you to blindly trust a black box. Each generated development is documented, versioned, and submitted for your approval before any deployment. You keep control; you only give up the dependence on the day rate.

In 2026, developing in X++ without an expensive consultant is no longer a myth. It is an operational reality accessible to any D365 CIO who chooses the right technology partner. The question is no longer whether it is possible — it is why you have not done it yet.