The goal of this page is not to dive into every detail of AI, but to provide a solid foundation and a starting point for anyone interested in learning about AI, whether you are a beginner or have some experience.

Table of Contents

History

Some fun history facts about AI, showing how long AI has been around and how it has evolved over time:

  • 1956: The term “artificial intelligence” was coined at the Dartmouth Conference.
  • 1966: ELIZA, an early natural language processing program, was created by Joseph Weizenbaum.
  • 1997: IBM’s Deep Blue defeated world chess champion Garry Kasparov.
  • 2011: IBM’s Watson won the quiz show Jeopardy! against human champions.
  • 2012: The breakthrough in deep learning with AlexNet winning the ImageNet competition.
  • 2014: The first Generative Adversarial Network (GAN) was introduced by Ian Goodfellow.
  • 2018: OpenAI released the first version of GPT (Generative Pre-trained Transformer).
  • 2020: OpenAI released GPT-3, a significant advancement in natural language processing.
  • 2023: OpenAI released GPT-4, further enhancing capabilities in language understanding and generation.

ML vs AI vs GenAI

  • What is ML?
  • What is AI and how is it different from ML?

  • What is GenAI and how is it different from AI?

  • What does HITL stand for and what does it mean in the context of AI? (Human In The Loop)
  • Scaled GenAI: Techniques and strategies for scaling generative AI models and applications.

GenAI part 1

Introduction

  • What is natural language processing (NLP)?
  • What is a large language model (LLM)?
  • What is a small language model (SLM)?
  • Are there other types of models?

Vendors

Many vendors, such as:

  • OpenAI
  • Google
  • Anthropic
  • Microsoft
  • Hugging Face

Others? What are the differences between them? What does a vendor do (bridge to models and providers)?

Models

What is a model?

There are many differences between models, like the following but more as well (described in the advanced section):

  • Training date
  • Cut-off
  • Large or small
  • Vendor

There are also many different models:

  • ChatGPT 3.5, 4.0, 4.1, 4.5 (LLM)
  • DeepSearch (LLM)
  • Gemini 2.5 Flash & Pro (LLM)
  • Claude Sonnet 4.0, Opus 4 (LLM)
  • Grok (LLM)
  • Phi (SLM)
  • MAI (Future AI model from Microsoft, LLM? Should maybe be placed in the future section along with other future models?)
  • DALL-E (Diffusion?)
  • Stable Diffusion (Diffusion?)
  • Midjourney (Diffusion?)
  • Imagen (Diffusion?)
  • Whisper (Speech recognition, what is this type of model called?)
  • AIVA (Music generation, what is this type of model called?)
  • GPT-4o & 4o mini (Multi modal? Large & small?)
  • Smol (SLM?)

FAQ

  • What is a GPT and why are not all models GPT?
  • GitHub Models is a suite of developer tools that take you from AI idea to ship, including a model catalog, prompt management, and quantitative evaluations. See providers section for more details.
  • What does multi-modal mean and how can I use it?
  • Why not train a model every month or week to keep it up-to-date? (link to training section in advanced?)

Providers

What are they?

(Technical) components of a hosted model include:

  • Proxy
  • Load balancer
  • Content filter
  • What else?

Hosted solutions:

  • OpenAI
  • Google
  • Microsoft
    • Azure OpenAI
    • GitHub Models
  • Hugging Face
  • Many more

Self-hosting is possible too:

  • Docker
  • Olama
  • Hugging Face

FAQ

  • Do all hosted solutions use my data?
  • Where is my data stored?
  • Can I opt-out of data collection?
  • How does the above change if I use an OpenAI model via Azure OpenAI?
  • How does GitHub Models relate to GitHub Copilot?

Prompts & messages

What are prompts and messages?

Different types of prompts and messages:

  • User prompt
  • System prompt
  • Assistant message
    • Suggestions
    • Completions

Prompt engineering is the process of designing and optimizing prompts to improve the performance of AI models.

  • Single shot prompts
  • Few shot prompts
  • Zero shot prompts
  • Chain of thought prompts

Reusable prompts are prompts that can be reused across different interactions.

Tokens & Tokenization

  • What is a token?
  • How does tokenization work?
  • How many tokens are in a message?
  • Why is this important?
  • Can I switch between different tokenizers?
  • Token limits and what happens when I exceed them? What to do?
  • How does tokenization differ between models, for instance text vs images vs audio?

Costs

  • Context
  • Chat history
  • Prompts

Problems with models

  • Hallucinations
  • Input-poisoning
  • Jailbreaking
  • Why can’t I use a LLM to calculate? Or count?

Fine-tuning a model

  • Grounding
  • Temperature
  • Top P

Advanced concepts

What is a neural network?

Training a model:

  • Transformers
  • Data
  • Cut-off date
  • Weights
  • Parameters
  • Context
  • Input size
  • Output size
  • Seed
  • Vocabulary
  • Attention

A deepdive into the above can be seen here where Andrej Karpathy builds ChatGPT from scratch.

These 3 might belong in the basics?

  • Vectors
  • Embeddings
  • Inference

GenAI part 2

Function calling

What is it?

Description: Vliegduur van A naar B
Definition:
  Params
    A: bla
    B: bla
Example:

How does the model match the function to the prompt?

Model Context Protocol (MCP)

Retrieval Augmented Generation (RAG)

  • What is it and why is it important?
  • How does it differ from MCP and function calling?
  • When to use or implement functions, when MCP and when RAG?

Agents & Agentic AI

  • What makes something an agent?
  • Is there a formal definition or interface?
  • What’s the difference compared to a MCP server and what to place where?
  • What does agentic mean?

Inspiration

Cool, but don’t know what to start with? Here are some examples to get you inspired.

Chat interfaces

Explain the popular chat interfaces and what you can do with them.

Office 365

What is the role of AI in Office 365?

Windows

GitHub

Other real-world examples

Show examples of how other companies are using AI in their applications.

Integrating AI into your applications

Tools and IDEs

  • Visual Studio (Code)
  • Rider
  • Claude Desktop?
  • GitHub Codespaces?
  • Jupyter Notebooks?

Copilot

  • GitHub Copilot, Microsoft Copilot, Azure Copilot, etc.
  • Ask vs Edit vs Agent in Copilot
  • Copilot spaces
  • Copilot coding agent (Padawan?)
  • Copilot review agent

Lots of examples can be found in the GitHub Copilot Hub.

FAQ

  • When should I refactor using my IDE and when should I use a tool like Copilot?

Azure AI services

What is Azure AI Foundry?

What is Azure OpenAI?

What other Azure services are there?

Do I need to use those “low-level” services directly anymore? Or can I skip ahead to the next section?

  • Azure AI Agent Service: Combine generative AI models with tools that allow agents to access and interact with real-world data sources.
  • Azure AI Foundry Models (previously Model Inference): Performs model inference for flagship models in the Azure AI model catalog.
  • Azure AI Search: Bring AI-powered cloud search to your mobile and web apps.
  • Bot Service: Create bots and connect them across multiple channels.
  • Content Safety: Detects unwanted or harmful content.
  • Custom Vision: Customize image recognition for your business needs.
  • Document Intelligence: Turn documents into intelligent, data-driven solutions.
  • Face: Detect and identify people and emotions in images.
  • Immersive Reader: Help users read and comprehend text more easily.
  • Language: Build apps with advanced natural language understanding capabilities.
  • Speech: Speech to text, text to speech, translation, and speaker recognition.
  • Translator: Use AI-powered translation technology to translate more than 100 languages and dialects.
  • Video Indexer: Extract actionable insights from your videos.
  • Vision: Analyze content in images and videos.

Languages & SDKs

Most popular languages have an SDK for AI, such as:

  • Python
  • JavaScript
  • Java
  • C#

Many SDKs:

  • Azure AI
  • OpenAI
  • Hugging Face
  • MCP
  • Smolagents: A library that enables you to run powerful agents in a few lines of code.
  • LangChain: A framework for building applications powered by language models.
  • LangGraph: A framework for building AI applications with a focus on graph-based workflows.
  • Semantic Kernel: Very important in the .NET space, so has a separate section.
  • Smart components: .NET Smart Components shows you how to add genuinely useful AI-powered features to your .NET apps quickly, easily, and without risking wasted effort.

Maybe show a matrix of currently popular languages and their SDKs?

Semantic Kernel

  • What is it?
  • What components are there and which AI capability do they deliver?
    • Plugins/planners vs (OpenAI?) function calling vs MCP?

FAQ

  • Adding Azure OpenAI vs OpenAI in C#, what are the differences?
  • What is still missing?

(Near) future

Lots of things are happening in the AI space, a few things to keep an eye on:

  • A2A (Agent to Agent)
  • ACP (Agent Communication Protocol) from BeeAI

A good starting point is here: What does MCP, A2A, and ACP mean?

Also SLIM (Secure Low-Latency Interactive Messaging), previously called AGP (Agent Gateway Protocol) is interesting. Find more information here: SLIM.

llms.txt is another initative to keep an eye on, suggesting adding a markdown file to websites to provide LLM-friendly content. Here is an example for GoFastMCP.

Microsoft is also embracing MCP and adding support for a central server registry, server isolation and other security features. Read the blogpost here.

Want to know more?

Show what Xebia Microsoft Services can do for companies.

Hugging Face is a great resource for learning about AI and finding models, datasets, and more.

Rob Bos’ LinkedIn Learning Course on AI development with GitHub models

GitHub Skills has a few courses on AI and related topics.

Let’s build GPT: from scratch, in code, spelled out by Andrej Karpathy.