AI Powered SaaS Support Agent

Automatically classify and respond to support emails while creating tasks and notifying your team
Created by
Kuldip Parmar
Kuldip Parmar
Founder
Relay.app screenshot of: AI Powered SaaS Support Agent

Steps in this workflow

1
Gmail logo
Start working when new Email received
2
OpenAI logo
Classify the Email into 5 Categories
3
Paths
A
Bug
4
OpenAI logo
Classify Email into 2 subset Categories
5
Paths
A
Explaner
6
DeepSeek logo
Generate Email replay to send
7
Gmail logo
Draft replay of the Email
8
Slack logo
Notify team on Slack about this update

Automatically process incoming support emails by classifying them into bugs, feature requests, general questions, integration requests, or other categories. Here's how it works:

  • The workflow activates when an email is received from a specified address
  • AI analyzes the email content and routes it to the appropriate handling path
  • For well-explained issues, tasks are created in ClickUp and warm replies are sent
  • For unclear requests, draft emails are generated asking for more details
  • Your team is notified via Slack at each step to stay informed
1
Gmail logo
Start working when new Email received

Monitors your Gmail inbox for new emails and triggers the workflow when an email is received. The trigger can be configured to only activate for specific senders or first emails in conversations.

Relay.app screenshot of: Start working when new Email received
2
Classify the Email into 5 Categories

Uses AI to classify the incoming email into one of five categories: Bug, Feature Request, General Question, Integration Request, or Other. The classification determines which path the workflow will follow for appropriate handling.

Prompt used
You are an expert email classifier for a support system. Read the email carefully and classify it into one of the following categories based on its content: 1. Bug → If the email reports an issue, malfunction, or technical problem (e.g., “Slack connection fails”). 2. Feature Request → If the user is suggesting or asking for a new feature (e.g., “Please add Microsoft Teams”). 3. Reply Drafter - General Question → If the user is asking a general or frequently asked question (e.g., “Is Relay.app free?”). Please remember that if it is related to our app Relay.app then only consider in this step otherwise add it to Other category. 4. Integration Request → If the user wants to integrate their system or tool into Relay.app (e.g., “Can we connect our CRM?”). 5. Other → If the email doesn’t fit any of the above categories. **Output only the category name. No explanation, no extra words.**
Relay.app screenshot of: Classify the Email into 5 Categories
3
Paths
ABug
BFeature Requiest
CGeneral Question
DIntergration Request
EOther
This path will be taken when the rules below match.
Relay.app screenshot of: Bug
Steps in Path A
4
Classify Email into 2 subset Categories

For bug reports, further classifies the email as either "Bug Noter" (clearly explained with sufficient detail) or "Explainer" (vague and needs more information). This determines whether to create a task or request clarification.

Prompt used
You are an expert support analyst. You will read a bug report email and classify it into one of two categories: 1. Bug Noter → If the bug is clearly explained with enough detail to understand the problem. This includes things like: - What the user was doing - Where the error occurred - Error codes or messages - Context or steps they took - Any attached images or proof 2. Explainer → If the bug is vague, lacks detail, or doesn’t provide enough information to understand or reproduce the issue. Only output one word: Either `Bug Noter` or `Explainer` No explanation. No extra text.
Relay.app screenshot of: Classify Email into 2 subset Categories
5
Paths
AExplaner
BBug Noter
This path will be taken when the rules below match.
Relay.app screenshot of: Explaner
Steps in Path A
6
Generate Email replay to send

Generates a polite email response asking the sender to provide more specific details about their bug report. The AI creates a warm, professional message that thanks them for reporting and requests additional information like steps to reproduce or error codes.

Prompt used
You are a helpful and empathetic support assistant at a product company. Your task is to read an email that mentions a bug, but doesn’t provide enough detail. Then, draft a short and kind email asking the sender to explain the issue more clearly. The reply must: - Address the sender by name (if available) - Be warm, professional, and human - Thank them for reporting the bug - Kindly ask for more specific explanation (like steps, error code, or screenshot) - Be concise and readable in under 120 seconds - End with a warm closing (signed by Luftan) Do not include subject line or greeting redundancy. Output only the email body in proper structure.
Relay.app screenshot of: Generate Email replay to send
7
Gmail logo
Draft replay of the Email

Creates a draft reply in Gmail with the AI-generated clarification request. The draft is saved to your Gmail account so you can review and send it when ready.

Relay.app screenshot of: Draft replay of the Email
8
Slack logo
Notify team on Slack about this update

Sends a notification to your configured Slack channel informing the team that a client submitted an unclear bug report. The message includes the client name and a link to the draft email for review.

Relay.app screenshot of: Notify team on Slack about this update
Author picture (Kuldip Parmar)

At TrueVibess, we're a team of 25 focused on helping startups scale through AI and automation. But like any growing SaaS company, we hit a wall — too many support requests, too little time. Our human team was stretched thin handling repetitive tickets, missing critical follow-ups, and juggling requests across email, ClickUp, and Slack.

That’s why we built this support AI agent — not as an experiment, but as a necessity. It now acts as the first layer of our support team: reading, classifying, replying, and escalating where needed — without delay or fatigue.

Since launch, we’ve responded to 10x more tickets with half the effort, and we haven’t looked back. This AI doesn’t just replace a helpdesk — it upgrades it. And now, we’re excited to share it with the world.

Kuldip Parmar
Founder