The field of generative Artificial Intelligence is experiencing a paradigm shifttowards Agentic AI. Unlike traditional AI’s reactive, task-specific nature orgenerative AI’s content-generation capabilities, Agentic AI systems arecharacterized by their autonomy, goal-directed behavior, and capacity to learnand adapt. This represents a significant advancement in workflow automation,enabling systems to operate within defined boundaries to enhance efficiency andfoster innovation.
This evolution fundamentally alters the GenAI developer’s role. Instead ofprescribing precise instructions, GenAI developers become architects ofautonomous decision-making spaces. Their focus shifts to defining goals, settingboundaries, and providing tools, thereby empowering the AI to determine thespecific sequence of actions needed to achieve its objectives.
This workshop aims to provide practitioners with the architectural thinking andpractical skills necessary to navigate this new paradigm and constructintelligent, autonomous systems.
Learning Objectives
Agentic systems become powerful byproviding them with the ability to retrieve from external information, tools forinteracting with their environment, and providing them with memory of the pastconversations.
Upon completion, participants will be equipped to:
Architect Autonomous Systems: Move beyond simple scripts to master corearchitectural patterns—including ReAct, Tree of Thoughts, and Plan & Execute—tobuild agents capable of reasoning, reflection, and solving multistep problems.
Engineer Robust Tool-Use: Design and integrate tools that serve as the handsand eyes of an agent. Learn the principles of the Agent-Computer Interface (ACI)to ensure agents can reliably interact with APIs, databases, and other externalsystems.
Build with Production-Grade Frameworks: Gain hands-on experience withleading frameworks such as LangChain & LangGraphbut also low-code alternatives such as Dify.ai, covering thetrade-offs and practical considerations of each. We will translate theoreticalknowledge into practical skills by building a complete, real-world agenticworkflow.
Deploy with Confidence: Navigate the challenges of development andunderstandability, including techniques for rapid prototyping and stakeholderengagement.
Structure of the Workshop
Routing is a central building blockwithin agentic systems, allowing the system to dynamically decide which path tofollow.Part 1: Building-blocks & Architecture of Agentic AI Systems
This section establishes the conceptual foundation for agentic systems, coveringtheir core components and the sophisticated cognitive architectures that enablereasoning.
Fundamental Building Blocks: The workshop will detail the essential patternsfor composing agentic workflows such as
Orchestration: Coordinating multiple-possibly specialized-AI agentsto achieve a shared objective.
Parallelization: Executing multiple tasks concurrently to enhance systemefficiency, either by dividing a large task into independent subtasks(sectioning) or by having multiple agents perform the same task for robust,aggregated results (voting).
Routing: Intelligently delegating tasks to the most appropriatespecialized agent or routing the chatflow.
Evaluator-Optimizer Loops (Reflection): Establishing a self-correctionloop where a generator agent iteratively refines its output based on anevaluator agent’s feedback against defined criteria.
Advanced Reasoning Architectures: The curriculum then advances to thecognitive architectures that structure an agent’s thought process. Beyond basicagent design, advanced reasoning architectures are crucial for structuring anagent’s cognitive process. These include ReAct, which connects reasoningwith tool use to ground an agent’s thoughts and reduce errors. Tree ofThoughts allows agents to explore multiple reasoning paths simultaneously,mimicking human trial-and-error for complex problem-solving. For continuousimprovement without fine-tuning, Reflexion enables agents to learn from pastfailures. Finally, Plan & Execute decouples high-level strategic planningfrom low-level action execution, improving performance on long, multistep tasksby maintaining a coherent strategy.
ReAct, combining reasoning and acting, allowsenables the agent to retrieve further information on a task using tools, andanswering the query only once it feels ready to do so.This section focuses on the critical discipline of designing the interfacebetween an agent’s reasoning engine and its ability to act in the world. Toolsare what transform a reasoning model into a functional actor, allowing it toaccess real-time information, execute code, and interact with proprietaryenterprise systems.
The central theme of this section is the ACI. Just as a Human-Computer Interface(HCI) determines the usability of traditional software for a human, the ACIdetermines the effectiveness of tools for an AI agent. A poorly designed ACI isa primary source of agent failure.
Part 3: Hands-on Implementation with Popular Frameworks
This section is dedicated to translating theory into practice. Participants willbuild a complete agentic workflow applying the introduced concepts usingindustry-standard frameworks, gaining practical skills that are immediatelyapplicable in a professional context.
Navigating the Framework Ecosystem: The workshop will provide an overview ofthe popular frameworks, highlighting their core philosophies and ideal use casesto inform architectural decisions.
Hands-On Project: Participants will apply the concepts from Part 1 and Part2 by building an agentic system. The project will involve defining agentroles, implementing a reasoning architecture, creating custom toolswith robust ACIs and memory, and orchestrating the entire workflow. A keycomponent of this project will be the visualization of the agent’s structureand their decision-making process, allowing participants to see how the agentoperates in real-time.
Part 4: Practical Considerations for Real-World Deployment
The final part bridges the gap between a working prototype and a reliable,scalable, and trustworthy enterprise application.It addresses the pragmatic challenges that arise when deploying agentic AI ina business context.
Rapid Prototyping: The workshop will discuss tools that accelerate thedevelopment cycle, including low-code and visual interfaces. Such approachesallow for the rapid creation, management, and testing of agentic workflowsthrough drag-and-drop interfaces, enabling faster iteration and experimentation.
Participatory Design: Furthermore, such approaches can be used to includestakeholders directly in the engineering of the agents to ensure that they solvethe right problems and align with business values.
Prerequisites
The training extends on the use of Large Language Models and therefore requiresbasic knowledge on prompting such. However, we will do our best to providepointers and short introductions of the key concepts involved in each part suchthat everybody should be able to follow. More importantly, in order to fullybenefit from our practical exercises, participants should have a basicunderstanding of Python and feel comfortable writing smaller parts ofPython code.
