ExamplesCustomer Support Bot

Customer Support Bot

Build an AI-powered customer support bot for Telegram in under 5 minutes using Aerostack’s bot platform.

What You’ll Build

A bot that:

  • Answers customer questions using your documentation (via MCP tools)
  • Creates support tickets when it can’t resolve an issue
  • Hands off to a human agent for complex cases
  • Tracks resolution rates and response times

Prerequisites

Step 1: Use the Template

  1. Go to Dashboard → Bots → New Bot
  2. Select the Customer Support template
  3. The template pre-configures:
    • A system prompt optimized for support conversations
    • Suggested MCP servers for knowledge base search
    • A workflow with escalation logic

Step 2: Configure Your Bot

{
  "name": "Support Bot",
  "platform": "telegram",
  "platform_config": {
    "bot_token": "YOUR_TELEGRAM_BOT_TOKEN",
    "webhook_secret": "a-random-secret-string"
  }
}

Step 3: Add Your Knowledge Base

Add MCP servers to your workspace that give the bot access to your docs:

  • Knowledge Base MCP — Upload your FAQ, docs, or help articles
  • Ticket System MCP — Connect to your ticketing system (Jira, Linear, etc.)

Step 4: Activate

Click Activate in the dashboard. The bot registers its webhook with Telegram and starts accepting messages.

Step 5: Test

Send a message to your bot on Telegram:

User: How do I reset my password?
Bot: To reset your password, go to Settings → Security → Reset Password.
     You'll receive a verification code at your registered email.
     Need more help? I can create a support ticket for you.

How It Works

User Message (Telegram)
  → Aerostack webhook receives message
  → Bot engine loads conversation history
  → LLM analyzes message + available tools
  → If tool needed: calls MCP workspace (search docs, create ticket)
  → LLM generates response with tool results
  → Response sent back to Telegram

Customizing the Workflow

The template includes a visual workflow you can customize:

  1. Trigger → Message received
  2. LLM Call → Classify intent (question / complaint / feature request)
  3. Logic Node → Route based on intent
  4. MCP Tool → Search knowledge base
  5. Send Message → Reply with answer
  6. Action → If unresolved, create ticket + escalate

See the Workflow Builder guide for more details.

Cost

  • Free tier: ~100 conversations/month using the prepaid wallet
  • BYOK mode: Bring your own OpenAI/Anthropic key for zero platform cost
  • Typical cost: ~$0.01-0.05 per conversation (varies by LLM model)

Next Steps

OverviewGitHub PR Review Agent