Agentic AI — Autonomous Agents (CrewAI/AutoGen/LangGraph)
A comparative workshop across three leading agent frameworks — CrewAI, AutoGen, and LangGraph — covering multi-agent orchestration, role division between agents, and production rollout of autonomous AI agent systems.
An agent is more than a chatbot with tools
An autonomous AI agent plans, acts, and observes the results of its actions in a loop, instead of answering a question once. The training starts with this fundamental difference — how function calling and agent memory differ from a classic LLM application — before moving on to building systems made up of multiple cooperating agents.
Three frameworks, one use case
Instead of teaching a single tool, the workshop walks through the same business use case implemented in turn in CrewAI, AutoGen, and LangGraph — so participants can directly compare the role-based approach, conversational human-in-the-loop patterns, and modeling flow as a state graph. This comparison, missing when learning a single framework, is the foundation for making an informed tool choice for a specific project.
From prototype to production with guardrails
Agent autonomy without control is an operational risk — infinite loops, unexpected decisions, rising model call costs. The final module of the training focuses on designing guardrails, error handling, and monitoring a multi-agent system, so the move from demo to production doesn’t end in uncontrolled agent behavior.
Benefits
- Understand the architectural differences between CrewAI, AutoGen, and LangGraph
- Design a multi-agent system with role division and task flow between agents
- Build control and safety mechanisms for autonomous agents (guardrails)
- Deploy an agent system to production with monitoring and error handling
Who is this training for?
Prerequisites
- Basic Python knowledge
- Basic understanding of how large language models (LLMs) and prompting work
Training program
Agentic AI fundamentals — from chatbot to autonomous agent
- What makes an agent different from a classic LLM application — the plan-act-observe loop
- Tools and function calling as the foundation of agent behavior
- Agent memory: short-term, long-term, and shared across agents
- When a single agent is enough, and when you need a multi-agent system
CrewAI — roles, crews, and processes
- The role-based model in CrewAI and defining an agent crew
- Sequential and hierarchical processes in CrewAI
- Building your first crew of agents completing a shared task
- CrewAI's strengths and limitations in production projects
AutoGen and LangGraph — conversational and graph-based orchestration
- AutoGen: conversational agents and human-in-the-loop patterns
- LangGraph: modeling agent flow as a state graph
- Comparing the three frameworks: when to choose which — decision criteria
- Combining approaches — hybrid agent architectures
Production: safety, monitoring, and scaling
- Guardrails: limiting agent autonomy and controlling critical decisions
- Handling errors, infinite loops, and unexpected agent behavior
- Monitoring and observability of a multi-agent system in production
- Operating costs of agent systems and optimizing model call volume
Delivery Methods
Online
- Convenience of participating from anywhere
- Interactive live sessions with trainer
- Materials available for 30 days
- No travel costs
On-site
- Direct contact with trainer and group
- Intensive hands-on workshops
- Networking with other participants
- Full focus on learning
Frequently asked questions
How is this different from Agentic AI — Building Autonomous Agents and Agentic AI — Building Autonomous Agents (LangGraph, CrewAI) in our catalog?
Those courses focus on one or two frameworks. This training covers three leading frameworks — including AutoGen, absent from the other catalog entries — with a direct architectural comparison and selection criteria, and puts more emphasis on multi-agent orchestration and production rollout than on learning a single tool.
Which framework — CrewAI, AutoGen, or LangGraph — is the best?
There is no single best choice — CrewAI works well for clearly defined team roles, AutoGen for conversational and human-in-the-loop patterns, and LangGraph for complex flows that require precise state control. In the training we build the same use case in all three frameworks, so participants can compare them on the same example.
Are autonomous agents safe in a production environment?
They require deliberate guardrail design — constraints defining which actions the agent can perform independently and which require human approval. In the training we show patterns for controlling autonomy, error handling, and preventing infinite loops, which are common problems in early agent system rollouts.
How much does it cost to run a multi-agent system in production?
Cost depends on the number of model calls per task — a multi-agent system with several planning and execution iterations can generate far more calls than a single prompt. In the training we cover optimization techniques: matching models of different price points to different agent roles, caching, and limiting iteration count.
Request a quote
Funding Options
Check funding options for your company
Development Services Database
Up to 80% funding for SMEs from EU funds
Check availabilityNational Training Fund
Up to 100% funding for employers
Learn moreTrusted by
We train teams at Poland's largest companies
Interested in this training?
Contact us - we'll prepare an offer tailored to your organization's needs.