Introduction
Unit 1 of 8 · Module 4: Develop generative AI apps that use tools
Read this lesson
The knowledge boundary
Welcome to Module 4, where your app learns to use tools. A language model is brilliant with words, but it's stuck inside a knowledge boundary: it can only reason about what was in its training data, like a well-read friend who hasn't seen a newspaper since the day they finished studying. Giving it tools lets your app reach far past what the model could manage on its own. A quick naming note: tools in a prompt are a different thing from Foundry Tools, the Azure AI APIs you met in Module 1.
Why tools matter
Think of tools as the model's hands and eyes in the real world. They fetch real-time information the model never saw, like today's weather, a live stock price, or a reply from some API. They take actions, like sending an email, adding a row to a database, or kicking off a workflow. They ground answers in facts, pulling trusted information in, so the model guesses less. They plug into the systems you already run: your databases, your services, your business rules. And they let you chain steps into smart workflows, like check the order, then refund it, then email the customer. Without tools, the model can only talk. With tools, it can look, think, and act.
In this module
In this module, your client app describes the tools inside the prompts it sends to the model, so the app itself stays in charge of the tool setup, like a custom assistant living inside your own code. That's a first step toward agentic AI, where the model, its instructions, and its tools are saved together as a named agent, which you'll build in the next learning path, Develop AI agents on Azure.
Facts from Microsoft's documentation: Use the Azure OpenAI Responses API (CC BY 4.0), rewritten in Loop's own words and examples. Narration, animation, and quiz by Loop. Microsoft, Azure, and Microsoft Foundry are trademarks of Microsoft. Loop is independent and not affiliated with or endorsed by Microsoft.