IK Workshop: Agentic AI Accelerator

Agentic AI is no longer about prompting LLMs. It’s about designing autonomous systems that can perceive, reason, plan, act, and learn — exactly the capabilities outlined in modern agentic architectures.

This workshop is built for technical professionals ready to go beyond “LLM usage” and actually engineer production-grade agentic workflows using proven design patterns such as:

• ReAct-style reasoning loops
• Prompt-chained pipelines
• Routing-based dynamic decision systems
• Parallelized agent execution
• Reflection and self-critique cycles
• Multi-agent collaboration models

If you’ve been experimenting with AI tools, this session helps you cross the gap between experimentation and deployable intelligent systems.

Stop just calling APIs — learn how to architect the next generation of autonomous software.

Why Attend?

Companies are rapidly shifting from “AI as a helper” to AI as a teammate.Teams now expect engineers and product leaders to understand:

• How agents reason across tasks
• How memory, planning, and tool-use interconnect
• How to design safe, robust, multi-step workflows
• How to build modular agent architectures leveraging tools, APIs, and databases
  

This workshop gives you a high–signal, engineering-focused playbook for:

• Architecting autonomous workflows
• Implementing memory-backed agents
• Integrating decision-making with tool calling
• Designing multi-agent systems with coordination protocols
• Evaluating robustness, safety, and failure propagation
  

What You'll Learn & Do

You don’t just hear concepts — you build.

1. The State of Play

A clear technical breakdown of how LLMs and GenAI evolved from “micro-taskers” to “autonomous collaborators” .

2. Agentic AI vs. Traditional ML

Why agents differ fundamentally from ML pipelines — moving from prediction to goal-driven autonomous action .

3. Core System Components

A deep dive into the seven architectural pillars that power robust agents: perception, reasoning, planning, memory, action/tooling, communication, and learning.

4. Agentic Design Patterns

You’ll see (and implement) patterns used in real systems today:

• Prompt chaining for sequential logic.
• Routing for dynamic intent classification and tool selection.
• Parallelization for latency reduction.
• Reflection for self-evaluation and improvement.
• Tool-use orchestration
• Multi-agent coordination structures
  
  

5. Hands-On Lab 1 — Build a File-System Agent

Using Strands Agents SDK + Llama 3.2.

• File listing and querying
• Reading/writing files
• Reasoning + tool-use cycles
• JSON function calling
• Error handling

This lab helps you internalize perception → reasoning → action loops.

6. Hands-On Lab 2 — Build a Multi-Agent Job Finder

Using CrewAI

• Specialized agents (analyst, strategist, filter, explainer)
• Multi-agent coordination
• Memory-enabled retrieval
• Routing-based decision flow
• Evaluation of results

You'll walk away with a working multi-agent pipeline — directly transferable to enterprise use cases.

7. Shipping to Production

How to evaluate, safeguard, optimize, and deploy agents:

• Avoiding error propagation
• Building safety layers
• Deployment patterns & MCP integration

🎟️ Book your spot now Seats are limited!

Don’t just adapt to the AI revolution - lead it. Get ready to understand career upleveling options like never before.