Redazione RHC : 4 September 2025 07:52
Microsoft today introduced POML (Prompt Orchestration Markup Language), a new markup language for prompt orchestration and specifically designed to enable rapid and structured prototyping of large language models (LLMs).
POML aims to address the limitations of traditional prompt development—often characterized by lack of structure, complex data integration, and format sensitivity—by offering a modular, readable, and maintainable approach. However, its introduction has sparked lively debate: some see it as a step forward in prompt engineering, while others see it as nothing more than a “reimagining” of XML, with a complexity that could reduce its practical adoption.
Syntactically, POML is similar to HTML: it uses semantic tags such as , , , , and to break down complex prompts into reusable, clearly defined components. This allows developers to systematically organize prompts, incorporate heterogeneous data (text, tables, images), and manage output formatting through a separate CSS-like style, reducing the instability typical of layout-aware templates for prompts.
In addition to the language, Microsoft has introduced an ecosystem of supporting tools. The Visual Studio Code extension provides syntax highlighting, contextual autocompletion, real-time preview, and error diagnostics. Additionally, the SDKs for Node.js and Python allow POML to be integrated into existing workflows and LLM-based frameworks. A typical example is the combined use of , and to define multimodal tasks that include images and output requirements.
The developer community has had a mixed reception of POML. On the one hand, some appreciate its structured approach, its templating engine (with variables, loops, and conditions), and its ability to simplify the management of complex prompts. On the other, there is no shortage of criticism regarding its similarity to XML and the feeling that writing prompts turns into a real coding activity, resulting in a steeper learning curve. Some observers also believe that, with the increasing use of AI agents and tool invocations, the rigidity of prompts is now less relevant, thus reducing the real need for a language like POML.
Promising application scenarios include dynamic content generation, A/B testing of prompt formats, and multimodal prompt creation. For example, POML can be used to automatically generate reports from tabular data or to quickly experiment with different output layouts by simply varying style sheets. Microsoft emphasizes that the separation of content and presentation makes POML adaptable to different LLMs and helps improve its overall robustness.
As the open source community grows and the toolchain improves, POML could become a gold standard in prompt engineering, paving the way for more robust and scalable development practices in the generative AI industry.
Redazione