Enthu Magic powers Auto QA in Enthu, allowing AI to evaluate agent calls using your scorecards, similar to how a human QA would.

Auto QA in Enthu can be triggered in two different ways, and understanding this difference is critical to avoid confusion around submission, visibility, and workflow.

Quick answer (TL;DR)

  • Auto QA evaluates calls using scorecards

  • Enthu Magic is the engine behind Auto QA

  • Auto QA can be triggered:

    1. Manually on a call

    2. Automatically via sampling

  • Manually triggered Auto QA is auto-submitted

  • Auto-sampled Auto QA requires manual submission

  • Sampling only decides which calls are evaluated

1. What Auto QA is (and what it is not)

What Auto QA does

  • Evaluates calls using your scorecards

  • Produces:

    • Question-level results

    • Scores

    • QA feedback

What Auto QA does NOT do

  • It does not decide which calls are picked (sampling does)

  • It does not provide customer context insights (Gen AI Insights do)

  • It does not automatically finalize all evaluations

2. What Enthu Magic does

Enthu Magic is the AI capability that:

  • Reads call transcripts

  • Understands agent–customer interactions

  • Automatically fills scorecard questions

Enthu Magic enables Auto QA.
It is not a report or insight by itself.

3. Two ways Auto QA happens in Enthu

Auto QA can be created in two different flows.

A. Manual Auto QA (Enthu Magic triggered on a call)

What this means

A user manually triggers Auto QA by clicking Enthu Magic on the main call screen.

How it works

  1. User opens a call

  2. Clicks Enthu Magic

  3. AI evaluates the call using the scorecard

Key behavior

  • No sampling involved

  • Best for:

    • Spot checks

    • One-off reviews

    • Validating AI output

B. Auto QA via Auto Sampling

What this means

Calls are:

  • First selected by Auto QA Sampling

  • Then evaluated automatically by AI

How to find these calls

  1. Go to the Call Analysis screen

  2. Click on Auto QAed Calls

  3. You’ll see all calls that were auto-sampled and AI-evaluated

4. Important: submission behavior for auto-sampled Auto QA

For auto-sampled Auto QA calls:

  • AI fills the scorecard automatically

  • Evaluations remain in Draft

  • A QA or Manager must manually submit the evaluation

Why this is designed this way

  • Ensures human oversight

  • Prevents blind acceptance of AI evaluations

  • Keeps QA ownership with the team

Auto-sampled Auto QA is AI-assisted, not AI-finalized.

5. Why a call was not Auto QA’d

A call may not be Auto QA’d if:

  • It was not selected by sampling rules

  • The call duration is too short

  • Transcript is missing or incomplete

  • Auto QA is not enabled on the scorecard

  • AI processing is still in progress

Not being Auto QA’d does not mean the call is missing or deleted.

6. Enthu Magic is missing or showing an error

Possible reasons

  • Auto QA is not enabled for the scorecard

  • Workspace does not have Enthu Magic enabled

  • Transcript quality is insufficient

  • Temporary AI processing issue

What to do

  • Refresh after some time

  • Check transcript availability

  • Confirm scorecard configuration

7. Can Auto QA evaluations be edited?

Yes, based on permissions:

  • QMs and Admin can:

    • Edit scores

    • Add comments

    • Override AI responses