Building Agentic AI Systems

Packt via Coursera

Go to class Write review

Details

Go to class

Provider

Coursera
Pricing

Paid Course
Languages

English
Certificate

Paid Certificate Available
Duration & workload

18 hours 29 minutes
Sessions

On-Demand
Subtitles

English

Found in

Overview

Coursera Flash Sale

40% Off Coursera Plus for 3 Months!

Grab it

This course takes you on an in-depth journey into building intelligent, autonomous AI agents that can reason, plan, and adapt. You'll gain practical knowledge on the design and deployment of agentic systems using generative AI principles, ensuring your ability to create robust AI solutions for real-world applications. By following the course, you'll enhance your ability to design AI systems capable of operating autonomously in dynamic environments. You'll work on real-world examples that reinforce the practical application of advanced AI techniques such as reflection, introspection, and collaboration. What sets this course apart is its combination of theoretical learning and hands-on implementation. We focus not only on the technology behind AI agents but also on ethical considerations, safety, and transparency, which are critical in today’s rapidly evolving AI landscape. This course is ideal for AI developers, machine learning engineers, and software architects with a solid programming background. If you're experienced with Python and eager to expand your skills in autonomous AI, this course is for you.

Syllabus

Fundamentals of Generative AI

In this section, we explore autoregressive LLMs like GPT-3 and PaLM for text generation and encoder-only models like BERT for NLU tasks such as text classification and NER. We discuss domain-specific LLMs and their applications in AI agents, generative AI for content creation, and multimodal models for images, videos, and audio. The section highlights practical use cases in media, fashion, marketing, and customer support, emphasizing ethical considerations, data quality, and computational challenges. It provides insights into building efficient and responsible AI solutions through real-world examples and technical concepts like NLU, NER, and generative models.

Principles of Agentic Systems

In this section, we explore agentic systems, focusing on self-governance, autonomy, and intelligent agent characteristics. We examine architectures like deliberative and hybrid systems, along with multi-agent interactions in logistics and travel booking assistants. Key concepts include autonomy types, task decomposition, and coordination mechanisms. The section emphasizes practical applications in decision-making, supply chain optimization, and adaptive systems, providing insights into building autonomous agents with real-world relevance.

Essential Components of Intelligent Agents

In this section, we explore knowledge representation using semantic networks and logic, reasoning methods like deductive and inductive reasoning, and learning mechanisms such as reinforcement and transfer learning. We examine how intelligent agents can adapt, make decisions, and improve through experience, with a focus on practical applications in dynamic environments.

Reflection and Introspection in Agents

In this section, we explore how reflection and introspection enhance intelligent agents by enabling them to analyze their reasoning, adapt their behavior, and improve performance through self-monitoring. Key concepts include meta-reasoning, self-explanation, and self-modeling, with practical implementations using CrewAI and real-world applications in customer service, financial trading, and e-commerce.

Enabling Tool Use and Planning in Agents

In this section, we explore integrating tool use and planning algorithms to enhance agent capabilities, covering REST API, SQL, and pandas 2.x for practical implementation. Key concepts include tool selection, workflow design, and applying algorithms like HTN and A* to enable efficient, context-aware decision-making in real-world scenarios.

Exploring the Coordinator, Worker, and Delegator Approach

In this section, we explore the coordinator-worker-delegator (CWD) model for designing multi-agent systems, focusing on role-based agent design and structured communication. We examine how to assign specific tasks to agents, establish efficient collaboration, and implement protocols for real-world AI applications, emphasizing adaptability and system resilience.

Effective Agentic System Design Techniques

In this section, we explore techniques for designing agentic systems with structured prompts, environment modeling, and memory strategies to ensure consistent performance. Key concepts include state space representation, context management, and workflow patterns like sequential and parallel processing for real-world AI applications.

Building Trust in Generative AI Systems

In this section, we examine strategies for building trust in generative AI systems through transparency, explainability, and bias mitigation. Key concepts include implementing clear communication, managing uncertainty, and ensuring ethical development to enhance user confidence and responsible AI deployment.

Managing Safety and Ethical Considerations

In this section, we examine strategies for safe and responsible AI deployment, focusing on mitigating risks like bias, misinformation, and data privacy violations. Key concepts include ethical guidelines, policy-based governance frameworks, and role-based access control to ensure AI systems operate within defined ethical and safety boundaries.

Common Use Cases and Applications

In this section, we examine how LLM-based agents are revolutionizing automation and human-AI collaboration across creative, conversational, and decision-making domains. The content highlights practical applications using Python, SQL, and REST API, emphasizing their role in adaptive, goal-directed systems that enhance efficiency and interaction in real-world scenarios.

Conclusion and Future Outlook

In this section, we explore the design and implementation of agentic systems using C# and REST API, while analyzing AI limitations and the challenges of achieving artificial general intelligence (AGI). We focus on practical applications, scalability, and ethical considerations in real-world AI solutions, emphasizing the importance of adaptability, reasoning, and efficient data handling with tools like pandas 2.x.