Skip to content

SimranShaikh20/Support-Agent-Automation

Repository files navigation

🤖 Support Agent with Zapier

This project demonstrates how to build a Customer Support AI Agent using Zapier, Gmail, and WhatsApp.
The agent automatically replies to common learner queries about Data Science and Data Analytics Bootcamps by searching official brochures.
If the answer isn't found, the query is escalated to the support team via WhatsApp.


Agent link -

https://agents.zapier.com/copy/191d15d7-9e96-42b2-a1f4-d13a2dd113a2

📌 Problem Statement

Support teams often spend hours replying to repetitive learner questions like:

  • "What's the duration of the Data Science Bootcamp?"
  • "Is there a certificate after course completion?"

This causes:

  • Delayed responses ⏳
  • Increased workload 😓
  • Inconsistent communication 📝

💡 Solution

The Support Agent solves this by:

  • Reading learner queries from Gmail.
  • Searching official Bootcamp Brochures (PDFs) for answers.
  • Drafting professional replies in Gmail.
  • Escalating unresolved queries to the support team via WhatsApp.

This ensures:

  • Fast responses for learners.
  • Automation of repetitive tasks.
  • Consistent, on-brand communication.
  • Human-in-the-loop for edge cases.

⚙️ How It Works (Workflow)

  1. Trigger: A new email arrives in Gmail.
  2. Zapier AI Agent:
    • Determines if it's a learner query.
    • Searches the Data Science and Data Analytics brochures.
    • Generates a reply.
    • Assigns a confidence score (0–10).
  3. Decision:
    • If confidence ≥ 7 → Draft email reply in Gmail.
    • If confidence < 7 → WhatsApp notification to support team: "Response needs manual attention."
  4. Support Team Action:
    • Reviews Gmail draft → Sends final reply.
    • If WhatsApp notification is received → Handles query manually.

text

alt text

🛠️ Setup Instructions

1. Create the Agent in Zapier

  • Go to Zapier Agents Dashboard → Click New Agent.
  • Select Support Email Agent or create from scratch.

2. Connect Gmail

  • Go to Zapier → My Apps → Add Connection → Gmail.
  • Sign in with your Gmail account.
  • Link it to your agent under Integrations.

3. Connect WhatsApp

  • Go to Zapier → My Apps → Add Connection → WhatsApp Notifications.
  • Authenticate via WhatsApp Business API (or approved provider).

4. Attach Bootcamp Brochures

  • Upload DS_Bootcamp_brochure.pdf and DA_Bootcamp_brochure.pdf as datasources in your agent prompt.
:datasource[DS_Bootcamp_brochure.pdf]{...}
:datasource[DA_Bootcamp_brochure.pdf]{...}

5. Add the Agent Prompt

Use this structured workflow:

1. Determine if the email is a learner query.
2. Search Data Science & Data Analytics brochures.
3. If answer found → Generate professional reply:
   - Start with "Hey there"
   - Thank user for reaching out
   - Warm, learner-first, humble tone
   - End with "Thanks & Regards, Team Codebasics"
4. Assign confidence score:
   - 9–10 = Exact match
   - 7–8 = Good match
   - <7  = Needs manual review
5. Action:
   - If score ≥7 → Draft Gmail reply
   - If score <7 → Send WhatsApp notification

📊 Example Scenarios

Scenario 1: Question on Data Science Bootcamp duration

  • Found in DS brochure → Draft Gmail reply created.

Scenario 2: Question on Data Engineering Bootcamp

  • Not found in DS/DA brochures → WhatsApp notification sent to support.

text

text

text

✅ Benefits

  • Reduces response time – Learners get instant answers.
  • 🤖 Automates workflows – From email → answer → escalation.
  • 📝 Ensures accuracy – Replies only come from brochures.
  • 👩‍💼 Frees support team – Focus only on unique cases.
  • 📈 Scales effortlessly – Handles unlimited queries.

📂 Architecture Diagram

graph TD
    A[📧 New Email in Gmail] --> B{🤖 Zapier AI Agent}
    B --> C[📋 Determine if Learner Query]
    C --> D[🔍 Search Bootcamp Brochures]
    D --> E[DS_Bootcamp_brochure.pdf]
    D --> F[DA_Bootcamp_brochure.pdf]
    E --> G[🎯 Generate Response]
    F --> G
    G --> H[📊 Assign Confidence Score 0-10]
    H --> I{Score ≥ 7?}
    I -->|Yes| J[✅ Draft Email Reply in Gmail]
    I -->|No| K[📱 WhatsApp Notification to Support Team]
    J --> L[👥 Support Team Reviews & Sends]
    K --> M[🛠️ Manual Query Handling]
    
    style A fill:#e1f5fe
    style B fill:#f3e5f5
    style J fill:#e8f5e8
    style K fill:#fff3e0
    style L fill:#e8f5e8
    style M fill:#fff3e0
Loading

Workflow Components

Component Description
Gmail Trigger Monitors incoming emails automatically
AI Agent Processes queries using Zapier's AI capabilities
PDF Datasources Official bootcamp brochures as knowledge base
Confidence Scoring Ensures quality control (0-10 scale)
Auto-Reply Creates draft responses for high-confidence answers
Escalation System WhatsApp notifications for manual review

text

🚀 Final Thoughts

This project shows how AI + Zapier + No-Code Tools can transform customer support. It's simple to set up, reliable, and drastically reduces manual workload.

Inspired by Codebasics, this Support Agent proves how automation and AI can create a smarter learner experience.


🔖 Tags

AI Agents Zapier Customer Support No-Code Automation Gmail Integration WhatsApp Automation Codebasics

Releases

No releases published

Packages

 
 
 

Contributors