Instructor-led · Advanced
Production Agentic Systems: Multi-Agent, RAG and Governance
Take multi-agent systems and RAG to production.
Free learning path 4
Agentic Workflows & Multi-Agent Orchestration
This free course teaches you to design agentic workflows and orchestrate several AI agents as one governed process. You build a workflow, coordinate specialist agents, add human approval where decisions carry risk, route each step to the right model and monitor runs and cost, hands-on in VDF AI Networks.
- 5 lessons
- 2 h
- Beginner
- Free, self-paced
Outcomes
What you will be able to do
- Design an agentic workflow from a real business process
- Coordinate specialist agents so each does one job well
- Put human approval exactly where a decision carries risk
- Route each workflow step to a model that fits the task
- Monitor runs, budgets and failures once the workflow is live
Before you start
Who it is for and what you need
Automation engineers, process owners and AI engineers building multi-step AI processes.
- Access to a VDF AI deployment with VDF AI Networks enabled
- At least one working agent; the Build Enterprise AI Agents path covers this
The lessons
5 lessons, in order
Each lesson is a hands-on build in your own workspace. For every field and option, see the VDF AI Networks documentation.
- 1 How to Build an Agentic Workflow, Step by Step Build an agentic workflow in VDF AI Networks. Describe the steps, review the generated plan, turn your rules into a verification step, pin models and run it.
- 2 How to Orchestrate Multiple AI Agents in One Workflow Orchestrate several AI agents in VDF AI Networks. Run specialists in parallel, reuse tested agents from Agents Hub, combine their work and check every source.
- 3 How to Add Human-in-the-Loop Approval to an AI Workflow Add a human approval step to an agentic workflow in VDF AI. Make it ask only for high-priority cases, review with a comment and see what a rejection does.
- 4 How to Route Each Workflow Step to the Right Model Choose a model per step in an agentic workflow. Compare automatic routing with pinned models in VDF AI Networks, read the routing decisions and set guardrails.
- 5 How to Monitor and Control Agentic Workflows Monitor agentic workflows in VDF AI. Watch a run live, read each step's model, time and tokens, diagnose a failed run and keep the evidence every run leaves.
Take it to production
Multi-agent processes that touch real systems need approvals, budgets and a complete record of every run, on infrastructure you control.
FAQ
Frequently asked questions
What is the difference between an agent and a network?
An agent does one job. A network in VDF AI Networks is a workflow of steps, each run by an agent, a tool, a check or a person, joined in order; you design it, save versions of it and run it as one unit.
Do I need agents from Agents Hub for this path?
No. The first lesson builds every step on the fly from a plain-language description. The orchestration lesson reuses two agents from the agents path, so complete that path first if you want to follow it exactly.
How long do the workflow runs take?
With pinned models, the workflows in these lessons completed in 7 to 20 seconds. The same four-step workflow on automatic routing took 1 minute 16 seconds, and an approval adds however long the reviewer takes.
After this path
Keep going
Try another free path, or the course that follows this one. It is free for VDF AI customers and partners.
Free path
VDF AI Fundamentals
Start here: the platform, chat and answers you can check.
Free path
Build Enterprise AI Agents
Build, test and govern enterprise AI agents.
Free path
Private RAG Engineering
Build and test RAG on your own infrastructure.
Free path
Self-Hosting Basics
Plan and run self-hosted VDF AI, from portal to runbook.
Free path
AI Governance and EU AI Act Compliance
Register, classify and document AI for the EU AI Act.