Demo Part 3 Step 2 Routing Queue AI Recording

Import: part-3-step-2-routing-queue-ai-recording.json

Goal

Add recording as the next layer on top of the combined scenario:

• extension-originated calls get the trigger-side recording branch
• AI-assisted offline calls start recording before the bridge

Changes Compared with Step 1

Reused node	Change	Why
internal	recording is turned back on	this re-enables the trigger-side Recording branch for extension-originated calls
dial_ai	no longer goes straight to bridge_ai	the AI path now makes the recording decision explicit first
added recording_internal, recording_ai	recording now has dedicated branches on both extension and AI paths	recording becomes a first-class feature instead of a hidden side effect

Why This Works

• recording_internal uses the Recording branch from the extension trigger. That keeps recording ownership aligned with the trigger contract instead of burying it inside generic media nodes.
• recording_ai sits before bridge_ai, so the AI recording decision happens before the caller and AI leg are connected.
• no tools are added yet. That keeps this step focused only on recording semantics across both call classes.

How to Test

1. Import and activate the workflow.
2. Test an extension-originated outbound call and verify that recording_internal runs.
3. Drive the queue into the offline AI branch and verify that recording_ai runs before bridge_ai.

Expected Result

Flow	Recording behavior
extension-originated call	recording_internal responds with the configured output path
AI offline branch	recording_ai starts before the AI leg is bridged

What The Next Example Adds

Part 3 / Step 3 keeps this recording layer and adds the attached voice-agent tools.