A is for Artificial Intelligence (AI)

'I am inevitable' - Thanos. I think it's finally time I talk about the two most used letters of the alphabet in recent times. Love it or hate it, the introduction of Artificial Intelligence (AI) has caused an enormous shift not only in the cloud space but across much wider industries.

But what actually is AI? And more importantly for those of us navigating the Microsoft cloud ecosystem, how on earth do we make sense of all the different AI services Azure has to offer? If you’ve looked at the Azure portal recently and felt completely overwhelmed by the sheer number of AI options, you are absolutely not alone. For this entry, I want to strip away the complex jargon and break down exactly what these tools are, what they do, and how you can begin to wrap your head around them.

The AI Toolkit

Fundamentally, you can think of the Azure AI ecosystem as a massive toolkit. Just like you have different tools for different jobs in a physical workshop—you wouldn't use a sledgehammer to carefully hang a picture frame—Azure provides different services depending on what you might want to create within the AI space. That might be a highly complex custom machine learning model, a generative AI chatbot that can talk to your customers, or just a simple scanner that pulls text off a PDF receipt.

Now, notice I said ‘toolkit’ at the start of this paragraph? Well, that’s because you rarely use just one tool to build a complete solution. Microsoft has broken these services down into a few core areas, each designed for a different level of expertise and a different type of project.

Let me explain using some real-world examples to make things a bit clearer:

• Azure AI Services (The Pre-built Tools): Let's start with the most accessible ones. Imagine you are building an application and you just want it to be able to transcribe audio, read text from a scanned document, or translate a web page from English to French. You don't need to be a data scientist or train a massive AI model for that. You can just plug in one of these pre-built Azure AI Services. They are essentially 'off-the-shelf' capabilities that you can easily integrate. It’s the quickest way to add intelligent features without needing a PhD in machine learning.

• Azure OpenAI Service (The Conversationalist): This is exactly what it sounds like, and it’s probably the one generating the most buzz right now. It gives you direct access to OpenAI’s incredibly powerful language models (like the ones powering ChatGPT). However, instead of using the public web version, you are getting these models firmly wrapped in the security, privacy, and access control boundaries of your own Azure tenant. Your company data isn't being used to train the public models, which is a massive win for corporate security.

• Microsoft Foundry (The Workbench): Formerly known as Azure AI Studio, if you want to start blending tools together, this is your main workbench. It acts as a central hub where developers can explore, build, and evaluate AI agents. It gives you access to a massive catalogue of models—not just Microsoft or OpenAI ones, but a huge range of open-source options too. You can bring your own data, test different prompts, and see which model works best for your specific use case.

• Azure AI Search (The Filing Cabinet): If you are building an AI tool that needs to read and understand your company's private documents, AI Search is your intelligent filing cabinet. It indexes your data so the AI models can retrieve the exact right information incredibly quickly. Without a good search tool, your AI is essentially reading a massive textbook with no index or contents page.

• Azure Machine Learning (The Heavy Machinery): Let’s say you work for a data-heavy organisation that needs to predict future sales trends based on decades of highly specific historical data. Azure Machine Learning (AML) is the heavyweight environment. This is where your data scientists can train, test, tune, and deploy their own custom models completely from scratch. It is incredibly powerful, but it requires a lot more technical knowledge to use effectively.

How to Start Your AI Journey

Now, if you’ve recently started your journey into Azure, my best advice is to start small. The beauty of the cloud is that you don't need to buy a massive server to try these things out.

If you are completely new, I highly recommend starting with the pre-built Azure AI Services. If you have a free Azure account (or your existing trial subscription credits), deploy a simple Vision or Speech service, and see if you can get it to analyse a picture or read a short audio file. Microsoft Learn has some fantastic, completely free sandbox modules that will walk you through these exact steps without you having to spend a single penny.

Don't feel like you need to understand Azure Machine Learning or complex model training right out of the gate. For the vast majority of people—even experienced IT professionals—leveraging the pre-built services or the Azure OpenAI service is more than enough to build some genuinely incredible solutions.

Closing Thoughts

Getting started with AI in Azure can feel like stepping into a vast, ever-evolving landscape. It can be intimidating, but the key is consistency and curiosity. This isn’t something to cram over a weekend or a single evening.

While AI can seem like a completely different world to traditional IT, it fundamentally relies on the same Azure principles of resource groups, networking, and logical structure that we've already discussed in previous entries. Think of the Azure AI portfolio as an exciting expansion of your existing cloud skills, not a replacement for them.

Take your time, explore what interests you, and don’t be afraid to experiment in a safe test environment. The more you engage with these tools and get that hands-on exposure, the more confident and capable you’ll become in navigating this new AI-driven world.