Build AI and Agentic Pipelines with Apache Airflow 3
Build with Apache Airflow 3 through Common AI Provider, Azure-hosted GPT, event-brief analysis, a read-only conflict agent and a validated JSON review packet.
1h 6m 57s • Intermediate • English with subtitles • Full HD
Course Overview
Build AI and Agentic Pipelines with Apache Airflow 3
Build review-ready outputs with deliberate workflow control.
Use Apache Airflow 3 to combine deterministic tasks, LLM tasks and agents in an AI pipeline. The course develops a Globoticket event workflow involving information extraction, context evaluation and tool use.
Follow the flow from a promoter request and stored event briefs to a conflict investigation and a validated JSON review packet. Airflow retains workflow control as AI tasks contribute information to the process.
Apache Airflow 3: 6 connected pipeline areas
Study the role of each component and the movement of information through the course workflow.
Combine deterministic and AI tasks
Understand how AI extends a deterministic pipeline through information extraction, context evaluation and tool use. Keep the workflow structure under Airflow control while using these capabilities.
Configure the provider and extract event data
Configure the Common AI Provider and Azure-hosted GPT. Use an LLM to turn a Globoticket promoter request into typed event data for the pipeline.
Analyze stored briefs and branch
Analyze event briefs held in Azure Blob Storage and work with AI-driven branching. Examine how the information from those briefs contributes to the workflow.
Investigate conflicts with a read-only agent
Use a read-only agent to investigate possible event conflicts through PostgreSQL and the Globoticket API. The investigation gathers information within the course workflow.
Validate and publish the review packet
Follow validation before publishing a JSON review packet to Azure Blob Storage. Treat the resulting packet as a review-ready output of the combined pipeline.
Inspect execution and usage
Inspect execution and usage in Apache Airflow 3 across the deterministic, LLM and agent components. Review how the tasks work together as the pipeline produces its output.
Explore the course details
The Globoticket request-to-review workflow
The Apache Airflow 3 workflow uses a Globoticket promoter request and event briefs as its business context. Typed event data and brief analysis feed the pipeline, with AI-driven branching and a read-only investigation of possible conflicts.
The investigation uses PostgreSQL and the Globoticket API. Validation precedes publication of the JSON review packet to Azure Blob Storage.
Agent boundaries and workflow control
Apache Airflow 3 keeps control of the workflow while deterministic tasks, LLM tasks and agents contribute to the output. The course shows how these components can work together to produce a packet that is ready for review.
The conflict agent's database and API investigation is read-only. Publishing the packet is a separate pipeline step after validation. Execution and usage inspection provide a view of the components involved in the process.
Course format
- Duration and level
- 1 hour 6 minutes 57 seconds / Intermediate
- Language
- English, with subtitles available
- Video
- MP4 / H.264, 1920 x 1080
- Audio
- AAC, 44.1 kHz, 2 channels
Explore related Development courses or IT & Software courses. For product reference, see the official Common AI Provider documentation.


