Module 3: AI Orchestration with Kestra

In this module, we learn how to orchestrate AI workflows using Kestra, an open-source orchestration platform. We start from the context problem that makes generic AI assistants unreliable, and end with autonomous multi-agent systems that can research, reason, and act without a fixed predetermined sequence of steps.

This module is created and taught by Will Russell from Kestra. Big thanks to Will and the Kestra team for making it possible.

Flows: flows/

Each lesson has its own video link at the top. If you'd like to watch them all in order, the full playlist is here.

Part 1: Using AI in Workflows

The first three lessons cover the theory behind using AI reliably in workflows — why context matters, why generic AI assistants fail, and how to set up your environment.

1. Introduction - Learning objectives, prerequisites, and why AI for workflows
2. Context Engineering - Why generic AI assistants produce incorrect Kestra flows
3. Setting up Kestra - Kestra setup, API keys, and importing the example flows

Part 2: AI Copilot

Part 2 shows how to use Kestra's AI Copilot to generate and refine flows by describing your inputs and goal, rather than building each step manually.

1. AI Copilot - Generating and refining flows with Kestra's built-in AI Copilot

Part 3: RAG Workflows

Part 3 shows how to ground AI responses in real data using Retrieval Augmented Generation, so the model works from facts instead of guessing. For a deeper dive into RAG and vector search, see Module 2.

1. Retrieval Augmented Generation - Ingesting documents, creating embeddings, and querying with context

Part 4: Agentic Workflows

Part 4 introduces AI agents that make autonomous decisions, use tools, and collaborate in multi-agent systems to complete complex tasks.

1. [AI Agents](lessons/06-agents.md) - Autonomous task execution, available tools, and observability
2. Multi-Agent Systems - Specialized agents collaborating on complex tasks

Part 5: Best Practices

Part 5 covers what you need to know before going to production - cost, security, observability, and when to use each approach.

1. Best Practices - Cost, security, observability, and production readiness
2. Next Steps - Resources, further reading, and where to go from here

Homework

• Homework

Extra Resources

• Meet Kestra in 60 seconds — a quick overview of what Kestra is
• Kestra Overview — a broader look at the platform
• Kestra Beginner Tutorial (10 min) — hands-on intro for getting started

Notes

Cohort|2026 notes by Nitin Gupta

• Add your notes above this line