Jueves de Quack con Bruno Capuano

Key Concepts

• Microsoft Build: The annual developer conference focused on cloud, data, AI, and software engineering.
• GitHub Copilot & CLI: AI-powered coding assistants and command-line interfaces used to streamline development workflows.
• Ollama: An open-source tool for running Large Language Models (LLMs) locally on personal hardware.
• MCP (Model Context Protocol): A standard protocol for AI models to interact with external tools and data sources.
• Microsoft Student Ambassadors: A global program for students to learn, build, and lead technology communities.
• Azure AI Foundry: A platform for building, evaluating, and deploying AI agents with features like real-time audio and web search.
• Semantic Kernel: An SDK that integrates LLMs with conventional programming languages (C#, Python, Java).

---

1. Microsoft Build and Session Navigation

The video highlights Microsoft Build, emphasizing its value for developers. With over 300 sessions, the speaker demonstrates how to use the Build CLI and GitHub Copilot to filter relevant content based on a specific project's codebase.

• Methodology: By installing a Build plugin into the IDE, the AI analyzes the local repository (e.g., a .NET 10 application) and suggests specific sessions (e.g., BRK260) that align with the technologies used, such as local model execution or Windows development.
• Actionable Insight: Developers can use AI to parse large event catalogs, ensuring they only attend sessions that provide immediate value to their current projects.

2. Local AI Development with Ollama

The speaker showcases Ollama Monitor, a custom Windows system tray application designed to track the status of local LLMs.

• Technical Detail: Ollama allows users to run models like Qwen 3.6, Llama 3.2, and Gemma 4 locally. The monitor application uses the ollama ps and ollama ls commands to provide real-time visibility into which models are loaded in memory.
• Real-world Application: This is particularly useful for developers running automated tests that trigger local models, as it provides a visual notification when a model is active, helping to debug resource usage and model selection.

3. AI-Assisted Development Workflow

The speaker demonstrates a "live coding" session where GitHub Copilot is used to add features to the Ollama Monitor.

• Process:
  1. Requirement: The user asks Copilot to add Windows notifications for model state changes.
  2. Implementation: Copilot generates the necessary code for the NotificationService and updates the UI/Settings.
  3. Refinement: The user provides iterative feedback (e.g., "Add a Save button," "Improve UI layout") to refine the AI-generated code.
• Key Argument: AI is best used to handle "boring" or repetitive tasks, allowing developers to focus on high-level architectural challenges. The speaker emphasizes being "polite" and specific with prompts to get better results from the AI.

4. Microsoft Student Ambassadors Program

Francis and Cristian, both senior ambassadors from Bolivia, discuss the program's evolution.

• Key Changes: The program has moved away from the "Alpha/Beta/Gold" milestone structure to a more streamlined model.
• Benefits: Ambassadors gain access to the GitHub Student Developer Pack, Azure credits, and support for organizing local community events (e.g., catering, swag, and technical resources).
• Advice: Cristian highlights that the program is open to students over 18 enrolled in higher education. He recommends using the program to organize local meetups to bridge the gap in technology awareness in their respective regions.

5. Azure AI Foundry and Real-Time Audio

The session concludes with a demonstration of Azure AI Foundry, specifically the new Real-Time Audio capabilities.

• Functionality: Agents can now be configured with a "Voice Mode," allowing for natural, low-latency conversational interaction.
• Technical Integration: The speaker shows how these agents can be connected to tools like web search. While the demo showed some "hallucinations" (e.g., the AI reading out a URL instead of just the temperature), it highlighted the ease of integrating complex AI features into applications with minimal code.

---

Synthesis and Conclusion

The video serves as a practical guide for modern developers to integrate AI into their daily workflows. The main takeaways are:

1. Leverage AI for Discovery: Use AI tools to navigate large amounts of information, such as conference catalogs.
2. Local AI is Accessible: Tools like Ollama make running powerful models locally feasible, provided the hardware supports it.
3. Iterative AI Coding: Treat AI as a junior developer; provide clear, specific instructions and iterate on the output to build functional tools.
4. Community Engagement: Programs like the Microsoft Student Ambassadors are vital for professional growth and community building, offering resources that go beyond simple software licenses.