Let's Build a Goal-Oriented AI Agent Using Semantic Kernel

Imagine an AI not limited to answering individual questions or chatting, but actively working towards a goal you've assigned to it.

In this session, we'll explore the building of an AI Agent – an autonomous actor that can execute tasks and achieve objectives on your behalf.

Along the way we will demystify:

1. 🧠 LLMs - What is a Large Language Model (LLM)
2. 📚 Tokens - What is a token and what are its roles
3. 💡 Embeddings - What are embedding models and vectors and what can they do for us
4. 🎯 Prompts - Beyond the basics
5. ⚙️ Tools - How can these be created and accessed using Semantic Kernel
6. 🤖 Agents - Let's put all these concepts to work!

The end result will be the core (or perhaps 'kernel' 😉) of an AI Agent - your virtual coworker willing to handle tasks on your behalf without. It will be built in C# using the open source, cross-platform Semantic Kernel library.

This talk assumes user-level familiarity with LLMs like ChatGPT or Microsoft Copilot and basic prompting. Anything else will be explained.

This talk does NOT use the Semantic Kernel Agent Framework which is still pre-release.

Designed as a 60- to 90-minute presentation. Target audience is developers and other technical roles. While C# code will be used to illustrate some of the concepts, it is not necessary to be a C# developer to understand this talk.

Empowering AI Agents with Tools using MCP

AI chatbots respond to prompts, but AI agents are much more interesting because they can take meaningful action on your behalf when you empower them with tools. In this talk we explain the concept of a tool and show how to allow an AI agent tool access through the emerging Model Context Protocol (MCP) standard. We will also survey the growing marketplace of MCP servers and how you can integrate them. Finally, we will consider how you can expose your own functionality through the MCP server standard.

By the end of this talk you should understand what tools are, why AI agents benefit from them, what role MCP plays, what problems it solves, when to use it, and how to create your own MCP-compatible tool.

Human Language is the New UI: How This Is Possible

Remember the first time you typed a question into ChatGPT and it responded with an answer so clear it seemed like a human wrote it? This was an eye-opening moment. This ability to understand and generate language was vastly more effective than anything we'd seen before.

This breakthrough in language understanding propelled ChatGPT to an unprecedented 100 million users in just two months, the fastest adoption of any application in history.

This language understanding is fueling a change in human-computer interaction, powering everything from chatbots to AI Agents. It's why Microsoft CEO Satya Nadella says, 'Human language is the new UI layer.' But how is this possible?

But how is this possible?

This talk will give you a practical look at how human language can be converted into a mathematical representation. We'll develop an intuitive understanding, starting with a brief introduction to tokens and embedding models. This will build to the idea of "semantic" representation, which is a core concept. We won't go deep on the math, but rather we’ll focus on how vectors are used to represent the meaning of words and how we can calculate the similarity between them.

From Theory to Practice: Language Understanding in Action

Though language understanding is also part of Large Language Models (LLMs), we will see semantic representation isolated in a purpose-built web app that lets you interactively compare text snippets side-by-side to see their semantic similarity. You'll be able to try it yourself and see how it works across different human languages and even programming languages.

Then we'll take a peek inside Microsoft AI Foundry to show an example of where you can find the technology used by the web app.

We’ll close by connecting the dots to other popular AI and AI-adjacent technologies like semantic search, vector databases, retrieval augmented generation (RAG), coding assistants, and even the humble chatbot. Don’t worry if any of these terms are unfamiliar. We’ll focus on how the language understanding piece is relevant. This can help you appreciate when these other technologies might be useful.

Target Audience

This talk is for any curious student. If you've used ChatGPT (or a similar tool), you're ready for this talk; no other technical background is required.

Takeaways

You'll walk away with a clearer idea of how human language can be the new UI layer and how you can make use of that understanding.

Student-targeted version