Circulos AI

Support AI Skill

Email Auto Response

Automate email support responses to deflect common questions before ticket creation. Configure intelligent email parsing, knowledge base matching, and reply-based ticket creation for email-first support operations. Use when setting up email auto-response rules, configuring email-to-ticket workflows, designing deflection emails, or managing email support automation. Triggers on phrases like "email auto-response", "email deflection", "email-to-ticket", "auto-reply support", "email parsing", "smart email response", "email automation support", "inbox automation".

Email Auto-Response & Deflection

Intercept support emails and attempt resolution before creating a ticket.

Workflow

Email Auto-Response Setup

Trigger: New email support channel; quarterly deflection optimization; high email volume period:

  1. Email parsing setup: Configure SMTP/IMAP integration; set up email parsing rules (extract sender, subject, body, attachments, thread history); filter non-support emails (newsletter opt-outs, BCC forwards, automated replies).
  2. Intent classification: Train ML model on historical support emails (5,000+ examples); classify into support categories (billing, technical, account, feature request, complaint, spam); set confidence thresholds.
  3. Knowledge base matching: For each category, define KB article mapping (category → top 5 relevant articles); configure semantic search fallback; set minimum match confidence (0.80).
  4. Auto-response template design: Create personalized email templates with: greeting, detected issue acknowledgment, relevant article(s), call-to-action ("reply if this doesn't help"), signature; A/B test variations.
  5. Reply-to-ticket workflow: Configure 24-hour reply monitoring; auto-create ticket if customer replies; preserve full email thread as ticket history; route to appropriate queue.
  6. Spam and abuse protection: Auto-detect email spam, phishing, harassment; flag for human review; never auto-respond to flagged emails; maintain blocklist.
  7. Launch and monitor: Enable for low-risk categories first (account questions, general info); monitor deflection rate, false positive rate, customer satisfaction; expand to technical categories after validation.
  8. Continuous optimization: Weekly review of missed deflections; monthly model retraining with new email data; quarterly template optimization.

Email Auto-Response Configuration

EMAIL AUTO-RESPONSE — CONFIGURATION
======================================

Ingestion:
  Email address: [email protected] (primary)
  Secondary addresses: [email protected], [email protected]
  Parsing engine: Mailparser / Parseur / Custom
  Thread detection: Group emails by conversation thread (In-Reply-To headers)

Classification:
  Model: Custom ML model (trained on 10,000+ historical emails)
  Categories: billing, technical, account, feature_request, complaint, general, spam
  Confidence threshold: 0.85 for auto-response; 0.60–0.85 for human review; <0.60 for ticket creation

Auto-Response Rules:
  Rule 1: If category = "general" AND confidence > 0.90 → auto-respond with KB article
  Rule 2: If category = "account" AND confidence > 0.85 → auto-respond with KB article
  Rule 3: If category = "billing" AND confidence > 0.85 → auto-respond with KB article
  Rule 4: If category = "technical" AND confidence > 0.90 → auto-respond with troubleshooting guide
  Rule 5: If category = "complaint" → NEVER auto-respond; create ticket immediately (Priority: High)
  Rule 6: If category = "spam" → auto-delete (with 30-day quarantine)

Template Structure:
  Subject: Re: [Original subject]
  Body:
    Hi [Sender Name],

    Thanks for reaching out. Based on your email about [detected issue], we think this article might help:

    [Article Title] — [Brief summary]
    [Link]

    If this doesn't solve your issue, simply reply to this email and we'll create a support ticket for you.

    Best regards,
    [Company] Support Team
    [Contact details]
    [Help center link]

Reply Monitoring:
  Window: 24 hours
  Action on reply: Create ticket with full email thread
  Queue: Based on reply content re-classification
  Priority: Standard (unless complaint language detected)
  SLA: 4 hours first response

Email Deflection Analytics

EMAIL DEFLECTION PERFORMANCE METRICS
=======================================

Period: Last 30 days

Volume:
  Total incoming emails: 3,500
  Auto-responded: 1,225 (35%)
  Converted to tickets: 1,820 (52%)
  Spam filtered: 455 (13%)

Deflection:
  Resolved without reply: 850 (69% of auto-responded)
  Replied to auto-response: 375 (31% of auto-responded)
  Overall deflection rate: 24% (850 / 3,500)

By Category:
  General inquiries:  85% deflection rate (highest)
  Account questions:  65% deflection rate
  Billing questions:  45% deflection rate
  Technical issues:   30% deflection rate (lowest — complex issues)
  Complaints:         0% deflection rate (by design)

Quality:
  False positive rate: 3.2% (auto-responded to email that needed human)
  Customer satisfaction (auto-response): 3.8/5.0
  Customer satisfaction (ticket): 4.4/5.0
  Complaint rate: 0.8% (emails where customer complained about auto-response)

Optimization Opportunities:
  1. Add more specific billing articles (currently 45% deflection vs. target 60%)
  2. Improve technical troubleshooting guide (30% → target 45%)
  3. Add video response option for visual troubleshooting
  4. Personalize article selection based on customer plan tier

Edge Cases

Integration Points

Disclaimer: All rights reserved by Circulos AI. These skills are specifically designed for Claude Code, Claude Cowork, Codex, and OpenClaw. When using or referencing any skill, please provide proper attribution to Circulos AI.