AI Workflows: Practical Systems with LLMs
This 2-day hands-on course explores how to build intelligent workflows and assistants by combining Large Language Models with structured logic and automation platforms.
This workshop introduces participants to the tooling and design principles needed to build AI-powered workflows and systems. We’ll cover LangFlow, n8n, LLM prompt engineering, fine-tuning, RAG pipelines, agent orchestration, and the trade-offs between hosted and local models. By the end of the course, you will have created working AI bots and automation tailored to your context.
Target audience and prior knowledge (Prerequisites):
- Developers
- DevOps Engineers
- Solution Architects
- Automation Engineers
- Data/AI Engineers
- Technical Product Owners
Teaching aids: Windows operating system computer. If desired, the participant can use their personal computer, which we ask you to inform the training company about.
Program:
Day 1 — Foundations: From Prompts to Workflows
- Understanding the AI Workflow Landscape: what LLMs can and can’t do, common workflow types, and modern tooling overview (LangFlow, n8n, LangChain)
- Creating your first LLM-powered automation: using n8n and LangFlow to build workflows and connect logic
- Prompt Engineering Essentials: few-shot prompting, roles, output formatting, common mistakes and fixes
- RAG in Practice: building a retrieval-augmented assistant using LlamaIndex or LangChain and a vector database
- Modular AI workflows: breaking down tasks into steps using multi-turn logic and orchestration patterns
Day 2 — Agents, Autonomy, and Evaluation
- Multi-Agent Systems Overview: MCP, A2A patterns, AutoGen-style orchestration, role assignment and structured dialogue
- Tool-Augmented Agents: implementing agents that call external tools, APIs, and internal services dynamically
- Evaluation and Testing: LangSmith, DeepEval, Promptfoo, designing test cases and using LLM-as-a-judge
- Local vs Cloud Models: performance, privacy, and cost trade-offs, including Ollama and fallback strategies
- Final Lab: design and present your own AI workflow or assistant using techniques and tools covered in the course
Learning methods: You can participate in the study joining the training through the online environment Zoom. The volume of training is 16 hours incl. 8 hours of practical exercises in the training environment.
Assessment method: The achievement of learning outcomes is assessed on the basis of feedback-based practical exercises carried out during the training.
Graduation Criteria: A graduate of the training receives a certificate if he performs all the practical exercises given during the training. Participants who have not achieved the learning outcomes will be issued a certificate of participation in the training upon request.
The price includes:
- Unique content – no fluff, only real-world AI and automation.
- Certificate of attendance – includes digital proof of your participation.
- Optional exam – test your knowledge and include your score on the certificate.
- Online format – attend from anywhere with full access to instructors.
- Recordings available – replay sessions at your own pace.
- Live and interactive – ask questions, discuss problems, get feedback.
- Q&A session each day – dig deeper into your edge cases and ideas.
- Quizzes and polls – keep the energy high and check understanding.
- Hands-on labs – build working flows with LangFlow, agents, RAG.
- Homework and feedback – optional exercises to reinforce your learning.
- Lifetime support – follow-up help and project advice anytime.
Curriculum group: 0613 Software and Application Development and Analysis.