Design AI Agentic Workflows

Stop experimenting with prompts. Start building systematic agent workflows.

Discover how to orchestrate multiple agents, integrate with external systems using MCP and Computer Use, implement quality verification, and deploy with proper security and monitoring.

Use pre-built templates and simulation tools to practice without API costs, and easily start building systematic agent workflows.

Why Attend

This intensive hands-on program takes you from foundational patterns to production deployment in 4 focused sessions.

Whether you need to automate customer support, build code review agents, or create research assistants, you’ll leave with working implementations and the patterns to build more.

Who Should Attend

Software Developers, Software Designers, Software Architects, and anyone interested in designing and building agentic AI workflows.

Practical Details

Language: English
Experiential, hands-on, interactive online learning with practical examples
Participants will be awarded a Mozaic Works certificate of completion

Duration

Format

Time

Start Date

Quantity:

- +

Price

Agenda

Building Blocks of Agentic Workflows
- Agents – Autonomous components with reasoning
- Memory – Short term, working, and long term
- Planning
- Coordination
- External Systems
- Verification
- Observability and Monitoring
Execution Patterns (inside the sandbox)
- Reasoning and Acting (ReAct)
- Reflection or Self Critique
- Generate and Test
- Direct Execution
Integration Patterns (outside the sandbox)
- Read only and Read-Write Integration
- Retrieval Augmented Generation (RAG) for agent memory
- External Tool Selection
- Interaction modalities
- Safety Controls
- Model Context Protocol (MCP)
Coordination Patterns
- Sequential Handoff
- Parallel Dispatch and Merge
- Conditional Routing
- Hierarchical Delegation
- Iterative Refinement Loop
- Consensus and Voting
Planning Patterns
- Hierarchical Decomposition
- Plan Then Execute
- Reactive Replanning
- Template and RAG-augmented Planning
- Milestone based Planning
Verification Patterns
- Generator – Critic
- Iterative Refinement Loop
- Multi Agent Consensus or Voting
- Specialized Multi-Agent Validation
- Human in the Loop
Testing and Evaluation
- Testing Non-Deterministic Systems
- Progressive Deployment as Testing Strategy
- Golden Set Technique
- Property Based Tests
- Manual Spot Checks
- LLM as Judge
- Multi-Turn Evaluation
- Record and Replay
Observability and Monitoring
- Metrics
- Specialized Tools for Agentic Workflows
- Tracing for Debugging
Securing Agentic Workflows
- Prompt Injection Vulnerability
- Model Poison Vulnerability
- Tool Poisoning Vulnerability
- Defense Practice: Mediate Actions
- Defense Practice: Limit Blast Radius
Hands-On Labs 🖥️
- Build Single Agent 🖥️
- Build a Multi-Agent Workflow 🖥️
- Secure an Agentic Workflow 🖥️
- Test an Agentic Workflow 🖥️

More details and full agenda here

Meet Your Trainers

Alex Bolboaca

CTO, Senior Trainer at Mozaic Works

With over 25 years of international experience in the world of software and technology, Alexandru shares about himself: Being deeply passionate about technology creation I am a continuous learner in the field. Between reading, trying out technologies and multiple programming languages, presenting and participating to developer communities and conferences, I have learned enough to write two books and teach hundreds of developers and architects. I love working with smart people who want to get better at what they do, whether it’s team work, communication, management or technical skills. I offer you my passion, my experience and my capacity to adapt to your needs. About Alex

More about Alex Bolboaca >

Adrian Bolboaca

VP of Innovation, Senior Trainer at Mozaic Works

With over 20 years of international experience in the world of software and transformations, Adrian shares about himself: I want to help my clients improve their business. I was helped by my professors, colleagues and mentors to continuously improve. Whenever I write code I focus on reducing domain complexity to obtain robust, maintainable software. The continuous improvement paradigm is important for the people I work with and for myself; this is why I like helping people, teams and businesses to become more adaptable and reach their goals. About Adi

More about Adrian Bolboaca >