Images by editor | chatgpt
# introduction
Sometimes, Agent AI is an AI that has taken the improvisation class and now it feels like it doesn't stop making its own decisions. Trying to define agent AI more accurately can feel like explaining jazz to someone who has never heard of music. It is guaranteed to bring some autonomy, some orchestration, and the people who actually run the show.
Well, there's no need to be confused by Agent AI anymore. This videowhich was recently recorded from ODSC talk and has become widely available to its creators, is comprehensive. 4-hour Agent AI Engineering WorkshopHosted by John Clone Jon Krohn YouTube Channel and Super Data Science Podcast, and Edward Donnerco-founder and CTO of Nebra.
The video jumps into Definition, design principles and development of AI agentshighlights unprecedented opportunities to derive business value from AI applications from 2025 onwards. Covering a variety of frameworks and practical applications, we show you how Language Model (LLM) output can control complex workflows and achieve task autonomy. Instructors highlight the rapid advances in LLM functionality and the possibility that agent systems can enhance or fully automate business processes.
The workshop highlights that Practical nature It comes with a Github repository with content attachments, and includes all the code that viewers can replicate and experiment with. Instructors frequently emphasize the rapid evolution of the field and the importance of starting small with agent projects to ensure success.
# What's covered?
Below are more specific topics explained in the video:
- Agent definition: In video, we define AI agents as programs that LLM control complex workflows, highlighting autonomy and appropriately distinguishing between simpler predefined workflows and dynamic agents.
- For Agent AI: It highlights the unprecedented opportunity for 2025 to derive business value from agent workflows, focusing on the rapid improvements in LLMS and its dramatic impact on benchmarks like Humanity's Last Examination (HLE) when used within the agent framework.
- Basic elementsCore concepts such as tools (enable LLMs to take actions) are explained along with inherent risks such as unpredictability and costs, as well as monitoring and guardrail strategies to mitigate them.
- What does Agent AI mean?: This workshop also covers the implications of agent AI, including workforce changes and strategies for future hard work careers in data science, highlighting skills such as multi-agent orchestration and basic knowledge.
Agent AI framework, the tools of the Agent Revolution, are covered.
- Model Context Protocol (MCP): An open source standard protocol for connecting agents with data sources and tools. It is often compared to “USBC for agent applications.”
- Openai Agents SDKA lightweight, simple and flexible framework used for deep research
- KuruwaiA heavy framework designed specifically for multi-agent systems.
- More complex frameworks like Langgraph and Microsoft Autogen It is also mentioned
Finally, practical video coding exercises include:
- A practical demonstration includes recreating OpenAI's deep research capabilities using the OpenAI Agent SDK and showing how agents can perform web searches and generate reports
- The discussion of agent system design principles covers five workflow design patterns: prompt chaining, routing, parallelization, orchestrator worker and evaluator optimizer.
- The construction of an autonomous software engineering team with Crewai has been proven. Agents work together to write and test Python code, generate a user interface, and highlight Crewai's “battery included” feature.
- The final project will show how to use MCP to develop autonomous traders, and how agents can access real-time market data, leverage persistent knowledge graphs, and perform web searches to make simulated trading decisions.
# The expected takeaway
After watching this video, viewers can:
- Understand the basic concepts of AI agents, including definitions, core components such as tools and autonomy, and the distinction between constrained workflows and dynamic agent systems.
- Implement agent systems using popular frameworks like Openai and Crewai, practice multi-agent collaboration setups, and take advantage of unique features such as structured output and automatic code execution.
- Understand and apply Model Context Protocol (MCP) to seamlessly integrate a wide range of tools and resources into agent applications, including the ability to create simple custom MCP servers.
- Develop practical agent applications as demonstrated by deep research capabilities recreation and autonomous software engineering teams and construction of simulated trading agents.
- Through monitoring and guardrails, we recognize and mitigate risks associated with the deployment of agent systems, such as unpredictability and cost management.
If you're looking for resources to set up Agent AI for you and show you how to take advantage of the fast-growing technology in AI engineering exploits more than this year, check out this incredible video from Jon Krohn and Edward Donner.
Matthew Mayo (@mattmayo13) Get a Master's degree in Computer Science and a Graduate Diploma in Data Mining. As editor-in-chief of Kdnuggets & Statology and contributor to Machine Learning Mastery, Matthew aims to provide access to complex concepts of data science. His professional interests include exploring natural language processing, language models, machine learning algorithms, and emerging AI. He is driven by his mission to democratize the knowledge of the data science community. Matthew has been coding since he was six years old.
