In the past few years, AI has become an unavoidable topic in just about every conversation within the ERP enterprise technology space. There is an official unspoken race for everyone to bring the power of AI to their own enterprise business solutions. But what does that really mean, and more importantly, what type of effort does this entail? In the current AI market, there are a lot of different flavors of AI and machine learning models to choose from. In the not-too-distant past, training a machine learning model was an elaborate effort involving lots of coding and feature engineering from raw datasets, and more importantly, it was very time and labor-consuming.
What is Microsoft Azure’s AutoML?
One example of a solution to this problem is Microsoft’s Azure AutoML (Automated Machine Learning) Service. Originally announced in 2018, Microsoft’s AutoML simplifies the process of creating and training personal machine learning models. Put simply, creating a model has been streamlined using automation with a low-code/no-code interface. Azure AutoML supports the creation of multiple model types, including Classification (yes/no type of questions), Regression (outcome prediction, revenue forecasting), Image Classification (image labeling), and Natural Language Processing (including sentiment analysis), among others.
Advancements in AutoML and model training
If we use a high-level example of creating a model, imagine taking a dataset of around 10,000 records. This data could come from a static exported file or from an API. You would then split the dataset randomly into two halves – a standard practice in machine learning. Various third-party tools can help with randomizing and splitting datasets. The first half of the data is used to train a model, while the second half is used to test it.
Here is where Azure AutoML shines. During this process, Azure AutoML utilizes automated feature engineering and task-based algorithm selection to evaluate the data against multiple algorithms, choosing the top-performing models. Additionally, automated hyperparameter tuning and ensemble modeling have recently been introduced to optimize performance further. What used to take hours or even days can now be accomplished in under 15 minutes for a dataset of this size.
Once a model is finalized, it is deployed as a RESTful API endpoint in Azure, allowing seamless integration with enterprise applications. Users can send input data via standard HTTP requests (such as POST in JSON format) and receive real-time predictions. From a security standpoint, it is important to note that NONE of this data is exposed to external databases or AI systems beyond the secured RESTful endpoint using OAuth 2.0 authentication.
The shift towards MLOps and AI lifecycle management
As AI adoption grows, enterprises must consider MLOps (Machine Learning Operations), which ensure AI models remain accurate, reliable, and up to date. Azure Machine Learning now offers Managed Endpoints as well, allowing enterprises to monitor model performance, automate retraining, and scale AI solutions efficiently.
This is particularly relevant for ERP and ESM platforms, where continuous model updates ensure real-time business insights, process automation, and AI-driven decision-making. Many leading platforms, including ServiceNow, Microsoft Dynamics 365, SAP, Oracle ERP Cloud, and Salesforce, have embraced AI as a core component of their ecosystems.
Another major trend is the rise of AI-powered copilots, which extend beyond predictive analytics into intelligent process automation. Microsoft’s Copilot for Dynamics 365 and Power Platform is an example of this shift. These AI-powered assistants leverage Azure OpenAI Service, combining AutoML and generative AI to streamline ERP workflows. This trend is not limited to Microsoft; ServiceNow and SAP are also integrating natural language processing (NLP) and AI-driven automation into their platforms to improve user experience and efficiency.
The key to success for your AI initiatives
The AI enterprise market space continues to expand globally, and having a knowledgeable partner to navigate the landscape can be the difference between a deployment taking four years or four quarters or less. Azure AutoML is just one piece of the larger AI ecosystem, which now includes generative AI, agentic AI, copilots, and automated MLOps pipelines.
Enterprises looking to stay competitive must consider not only the ease of AI model creation but also how these models are deployed, monitored, and enhanced over time. As AI becomes an integral part of ERP and ESM solutions, the ability to seamlessly integrate AI-powered decision-making will define the next generation of enterprise business solutions.
With Rimini Street’s partnership with ServiceNow, Rimini Consult™, our professional services arm, can help you build those integrations and take full advantage of the capabilities of Azure AutoML and other Microsoft applications.
