Email and transactional content QA

Business value

• Improves trust in customer-facing communication.
• Catches vague operational emails before they cause support replies.
• Protects brand credibility in sensitive workflows.

Agent job to be done

Act as an email QA reviewer. Make sure each email is specific, useful, direct, and aligned with the customer’s actual context.

format: otherintended_use: publishdomain: email / transactional content

When to call VeracityAPI

Run before campaign send, automation activation, or high-volume transactional template deployment.

What text to submit

Subject, preview text, body, CTA, transactional variables with representative sample values, and audience/context notes.

Decision policy

• allow: low risk and no high-severity evidence.
• revise: medium risk, vague next steps, generic reassurance, or unsupported claims.
• human_review: high risk for refunds, safety warnings, billing, compliance, or account-access emails.
• local rule: transactional emails must state the exact event, next step, and support path.

Automation recipe

• Email agent drafts template.
• Render with representative variables.
• Score rendered output.
• If revise, patch vague spans and rescore.
• Only activate templates that pass.

Evidence spans agents should inspect

• generic reassurance without details
• unclear next steps
• boilerplate that ignores customer context
• unsupported safety or billing claims

Policy pseudocode

if (result.recommended_action === "allow") continueWorkflow();
if (result.recommended_action === "revise") rewriteWith(result.evidence, result.recommended_fixes);
if (result.recommended_action === "human_review") queueForHumanReview(result);
if (result.recommended_action === "reject") discardOrRebuild();

KPIs to track

• support replies per email sent
• template revise rate
• newsletter engagement
• refund/support escalation rate
• time to approve lifecycle emails

What can go wrong

• Short transactional emails may have lower confidence; combine with required-field checks.
• Do not submit secrets, tokens, or personal data unless privacy policy permits; use representative placeholders.
• Compliance/legal emails still need appropriate review.

Cost and latency notes

Analyze only is $0.005 per 1,000 characters; Analyze + revise with auto_revise=true is $0.010 per 1,000 characters. Both round up to the nearest 1,000 characters. Short captions/emails usually cost $0.005; longer pages or chapters scale linearly by length. Current v0.1 latency is LLM-bound, so batch/concurrent orchestration is recommended for high-volume pipelines.