klbr/n8nworkflows.xyz
1
0
Fork 0
mirror of https://github.com/khoaliber/n8nworkflows.xyz.git synced 2026-04-25 19:21:05 +00:00
Code Issues Packages Projects Releases Wiki Activity

creation

Browse Source
This commit is contained in:
nusquama
2025-11-12 19:34:19 +01:00
parent 9a3557b9d4
commit 42e17cfccc
1 changed files with 296 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
workflows/Automated Customer Feedback Triage with Gemini, Jotform, Google Sheets & Gmail Replies-9636/readme-9636.md
+296
View File
@@ -0,0 +1,296 @@
Automated Customer Feedback Triage with Gemini, Jotform, Google Sheets & Gmail Replies
https://n8nworkflows.xyz/workflows/automated-customer-feedback-triage-with-gemini--jotform--google-sheets---gmail-replies-9636
# Automated Customer Feedback Triage with Gemini, Jotform, Google Sheets & Gmail Replies
### 1. Workflow Overview
This workflow automates the triage and response process for customer feedback collected via a Jotform. It is designed to classify incoming feedback into categories (comments, questions, suggestions), analyze sentiment, and route or respond accordingly:
- **1.1 Input Reception:** Listens for new Jotform submissions containing customer feedback.
- **1.2 Feedback Classification:** Uses a Switch node to categorize feedback into comments, questions, or suggestions.
- **1.3 Sentiment Analysis & Logging for Comments:** Performs sentiment analysis on comments and logs them in a Google Sheet.
- **1.4 Question Handling:** Uses a LangChain Q&A agent backed by a Google Sheets knowledge base to generate friendly replies, which are emailed to the customer.
- **1.5 Suggestion Handling:** Summarizes suggestions, sends a Telegram alert, and logs suggestions in a Google Sheet.
- **1.6 Urgent Alerting:** If sentiment analysis detects angry feedback, sends an urgent Telegram notification to support.
The workflow integrates multiple services: Jotform, Google Sheets, Gmail, Telegram, and Google Gemini (Google Palm API) for AI-powered language processing.
---
### 2. Block-by-Block Analysis
#### 2.1 Input Reception
- **Overview:** Captures new customer feedback submissions from a configured Jotform.
- **Nodes Involved:** JotForm Trigger
- **Node Details:**
- **JotForm Trigger**
- Type: Trigger node specifically for Jotform.
- Configuration: Linked to a specific Jotform form (ID configured via credentials).
- Key variables: Feedback fields such as "Name", "E-mail", and "Describe Your Feedback:" are extracted.
- Input: Incoming webhook from Jotform submission.
- Output: Passes submission JSON to downstream nodes.
- Failure modes: Webhook failures, API credential errors, invalid form mapping.
- Version: 1
#### 2.2 Feedback Classification
- **Overview:** Routes feedback into categories "comments", "questions", or "suggestions" based on the "Feedback Type" field from the form.
- **Nodes Involved:** Switch
- **Node Details:**
- **Switch**
- Type: Conditional routing node.
- Configuration: Matches lowercase "Feedback Type" to one of three outputs: comments, questions, suggestions.
- Expressions: `{{$json['Feedback Type'].toLowerCase()}}` used for case-insensitive comparison.
- Inputs: From JotForm Trigger.
- Outputs: Three separate branches for each feedback type.
- Failure modes: Missing or malformed "Feedback Type" field causing no match.
- Version: 3.3
#### 2.3 Sentiment Analysis & Logging for Comments
- **Overview:** Processes feedback labeled as comments by analyzing sentiment and storing results in a Google Sheet. If sentiment is angry, sends Telegram alert.
- **Nodes Involved:** Google Gemini Chat Model → Sentiment Analysis → Store to Comments Sheet → Send to Support Group (conditional)
- **Node Details:**
- **Google Gemini Chat Model**
- Type: AI language model node (Google Gemini/Google Palm API).
- Configuration: Default options for language model.
- Role: Likely provides context or preprocessing for sentiment node.
- Inputs: From Switch node’s “comments” output.
- Outputs: To Sentiment Analysis.
- Failure modes: API authentication, rate limits.
- Version: 1
- **Sentiment Analysis**
- Type: LangChain sentiment analysis node.
- Configuration: Analyzes text from `{{$json['Describe Your Feedback:']}}`.
- Inputs: From Gemini model.
- Outputs: To Store to Comments Sheet and Send to Support Group (depending on sentiment).
- Failure modes: Empty text, API errors.
- Version: 1.1
- **Store to Comments Sheet**
- Type: Google Sheets append node.
- Configuration: Adds Name, Email, Comments, Sentiment, and current date to “comments” sheet.
- Inputs: Sentiment Analysis output.
- Outputs: None (terminal).
- Failure modes: Sheet access, permission errors.
- Version: 4.7
- **Send to Support Group**
- Type: Telegram message node.
- Configuration: Sends alert message with customer details and feedback if sentiment is angry/urgent.
- Inputs: Sentiment Analysis output (conditional).
- Outputs: To Store to Comments Sheet (duplicates? Possibly fallback).
- Failure modes: Telegram API limits, wrong chat ID.
- Version: 1.2
#### 2.4 Question Handling
- **Overview:** For feedback tagged as questions, queries a Google Sheets knowledge base using a LangChain Q&A agent to generate an answer, then sends a styled email reply.
- **Nodes Involved:** Read Database → QnA Agent → Reply Customer
- **Node Details:**
- **Read Database**
- Type: Google Sheets Tool node.
- Configuration: Reads FAQ/knowledge base sheet by document ID and sheet name.
- Inputs: None (triggered via downstream “ai_tool” input from QnA Agent).
- Outputs: Data for QnA Agent.
- Failure modes: Sheet access errors, invalid document or sheet IDs.
- Version: 4.7
- **QnA Agent**
- Type: LangChain agent node.
- Configuration: Uses feedback text as user prompt, system message sets role as helpful assistant with politeness constraints, fed with Google Sheets data.
- Expressions: `"user: {{ $json['Describe Your Feedback:'] }}"`.
- Inputs: From Switch “questions” output and Read Database.
- Outputs: To Reply Customer.
- Failure modes: Model API errors, prompt formatting issues.
- Version: 2.2
- **Reply Customer**
- Type: Gmail node.
- Configuration: Sends an HTML email reply with personalized content, including original question and AI-generated answer.
- Expressions: Uses Mustache-like template to inject name, email, question, and answer.
- Inputs: QnA Agent output.
- Outputs: Terminal node.
- Failure modes: Gmail OAuth issues, template rendering errors.
- Version: 2.1
#### 2.5 Suggestion Handling
- **Overview:** For suggestions, summarizes content with AI, alerts the product Telegram channel, and logs the suggestion in a dedicated Google Sheet.
- **Nodes Involved:** Google Gemini Chat Model1 → Summarize Suggestions → Send a text message → Add to Suggestions backlog
- **Node Details:**
- **Google Gemini Chat Model1**
- Type: Google Gemini language model.
- Configuration: Temperature set to 0.4 for balanced creativity.
- Inputs: Switch “suggestions” output.
- Outputs: To Summarize Suggestions and QnA Agent (appears in connections).
- Failure modes: API limits.
- Version: 1
- **Summarize Suggestions**
- Type: LangChain chain summarization node.
- Configuration: Summarizes suggestion text.
- Inputs: From Gemini Model1.
- Outputs: To Send a text message.
- Failure modes: Empty input, API errors.
- Version: 2.1
- **Send a text message**
- Type: Telegram node.
- Configuration: Sends summary and full suggestion text to product Telegram chat.
- Inputs: From Summarize Suggestions.
- Outputs: To Add to Suggestions backlog.
- Failure modes: Telegram API or chat ID issues.
- Version: 1.2
- **Add to Suggestions backlog**
- Type: Google Sheets append node.
- Configuration: Logs Name, Email, Summary, Suggestions, and Created Date to suggestions sheet.
- Inputs: From Send a text message.
- Outputs: Terminal.
- Failure modes: Sheet access, mapping errors.
- Version: 4.7
#### 2.6 General Logging for Comments
- **Overview:** Stores all comments along with sentiment and metadata to a Google Sheet for reporting and dashboards.
- **Nodes Involved:** Store to Comments Sheet (also involved in 2.3)
- **Node Details:**
- Same as detailed in 2.3.
---
### 3. Summary Table
| Node Name | Node Type | Functional Role | Input Node(s) | Output Node(s) | Sticky Note |
|-----------------------|----------------------------------|----------------------------------------------|-----------------------|----------------------------------|------------------------------------------------------------------------------------------------------------------------------------------------------------|
| JotForm Trigger | n8n-nodes-base.jotFormTrigger | Capture new customer feedback form submissions | (Trigger) | Switch | New submission from Jotform: starts when new entry arrives with mapped fields. |
| Switch | n8n-nodes-base.switch | Classify feedback into comments/questions/suggestions | JotForm Trigger | Sentiment Analysis, QnA Agent, Summarize Suggestions | |
| Google Gemini Chat Model | @n8n/n8n-nodes-langchain.lmChatGoogleGemini | Provide AI context/preprocessing for sentiment analysis | Switch (comments) | Sentiment Analysis | |
| Sentiment Analysis | @n8n/n8n-nodes-langchain.sentimentAnalysis | Analyze sentiment of comments | Google Gemini Chat Model | Store to Comments Sheet, Send to Support Group | Log for reporting: appends comments with sentiment and metadata. |
| Store to Comments Sheet | n8n-nodes-base.googleSheets | Log comments and sentiment into Google Sheet | Sentiment Analysis, Send to Support Group | None | |
| Send to Support Group | n8n-nodes-base.telegram | Alert support via Telegram on angry feedback | Sentiment Analysis | Store to Comments Sheet | |
| Read Database | n8n-nodes-base.googleSheetsTool | Load FAQ knowledge base for Q&A | (ai_tool input from QnA Agent) | QnA Agent | Answer with Q&A agent + email reply: loads Q&A sheet for grounding. |
| QnA Agent | @n8n/n8n-nodes-langchain.agent | Generate paraphrased, polite answers for questions | Switch (questions), Read Database | Reply Customer | |
| Reply Customer | n8n-nodes-base.gmail | Send styled HTML email replies to customers | QnA Agent | None | |
| Google Gemini Chat Model1 | @n8n/n8n-nodes-langchain.lmChatGoogleGemini | AI model for summarizing suggestions | Switch (suggestions) | Summarize Suggestions, QnA Agent | Summarize + alert + log: summarization and Telegram alert for suggestions. |
| Summarize Suggestions | @n8n/n8n-nodes-langchain.chainSummarization | Summarize suggestion text | Google Gemini Chat Model1 | Send a text message | |
| Send a text message | n8n-nodes-base.telegram | Notify product Telegram channel with summary and full suggestion | Summarize Suggestions | Add to Suggestions backlog | |
| Add to Suggestions backlog | n8n-nodes-base.googleSheets | Log suggestions and metadata into Google Sheet | Send a text message | None | |
| Sticky Note | n8n-nodes-base.stickyNote | Workflow description and overview | None | None | Jotform feedback triage & auto-reply system overview with setup tips. |
| Sticky Note1 | n8n-nodes-base.stickyNote | Description of Jotform trigger input | None | None | New submission from Jotform explanation. |
| Sticky Note2 | n8n-nodes-base.stickyNote | Explanation of Q&A agent and email reply process | None | None | Answer with Q&A agent + email reply details. |
| Sticky Note3 | n8n-nodes-base.stickyNote | Explanation of suggestions flow | None | None | Summarize + alert + log for suggestions. |
| Sticky Note4 | n8n-nodes-base.stickyNote | Explanation of comment logging | None | None | Log for reporting details. |
---
### 4. Reproducing the Workflow from Scratch
1. **Create a JotForm Trigger Node**
- Select type: JotForm Trigger.
- Configure with your Jotform API credentials.
- Select the specific form ID to listen for new submissions.
- Map form fields: ensure "Name", "E-mail", "Feedback Type", and "Describe Your Feedback:" are included.
2. **Add a Switch Node to Classify Feedback**
- Add a Switch node connected to JotForm Trigger.
- Configure three outputs with rules:
- Output "comments": when `{{$json["Feedback Type"].toLowerCase()}}` equals "comments"
- Output "questions": when equals "questions"
- Output "suggestions": when equals "suggestions"
- Use case-insensitive matching.
3. **Set Up Google Gemini Chat Model Node for Comments**
- Add Google Gemini Chat Model node connected to Switch “comments” output.
- Use default options for the AI model.
- Authenticate with Google Palm API credentials.
4. **Add Sentiment Analysis Node**
- Connect Google Gemini Chat Model node output to Sentiment Analysis node.
- Configure input text as `{{$json["Describe Your Feedback:"]}}`.
- Use default options.
- Version 1.1 or later recommended.
5. **Configure Store to Comments Sheet Node**
- Connect Sentiment Analysis node output to Google Sheets node.
- Set operation to "append".
- Map columns: Name, Email, Comments, Sentiment category, Created Date (format: yyyy-MM-dd).
- Provide Google Sheets OAuth2 credentials.
- Specify target spreadsheet ID and sheet name for comments.
6. **Set Up Conditional Telegram Alert for Angry Feedback**
- From Sentiment Analysis node, add Telegram node.
- Configure message template to include customer name, email, message, and urgent alert text.
- Use Telegram API credentials and specify target chat ID.
- Connect Telegram node output back to Store to Comments Sheet (optional redundancy).
7. **Configure Read Database Node for Q&A**
- Add Google Sheets Tool node.
- Configure to read your FAQ knowledge base sheet via document ID and sheet name.
- Authenticate with Google Sheets OAuth2.
- No direct trigger connection; will be triggered as AI tool input.
8. **Add QnA Agent Node**
- Connect Switch “questions” output to QnA Agent node.
- Set prompt text: `"user: {{ $json['Describe Your Feedback:'] }}"`.
- Define system message with role instructions and polite tone constraints.
- Link AI tool input from Read Database node.
- Authenticate with Google Palm API credentials.
9. **Add Gmail Reply Node**
- Connect QnA Agent output to Gmail node.
- Configure to send email to `{{$json['E-mail']}}`.
- Compose styled HTML message:
- Include customer name, original question, AI-generated answer.
- Set subject line including short snippet of feedback.
- Authenticate with Gmail OAuth2 credentials.
10. **Set Up Google Gemini Chat Model1 Node for Suggestions**
- Connect Switch “suggestions” output to Google Gemini Chat Model1 node.
- Set temperature to 0.4.
- Authenticate with Google Palm API.
11. **Add Summarize Suggestions Node**
- Connect Google Gemini Chat Model1 output to Summarize Suggestions node.
- Use default summarization options.
12. **Configure Telegram Notification for Suggestions**
- Connect Summarize Suggestions node to Telegram node.
- Compose message with customer name, email, summary, and full suggestion.
- Authenticate with Telegram API, specify target chat ID.
13. **Add Suggestions Logging Node**
- Connect Telegram node to Google Sheets append node.
- Map columns: Name, Email, Summary, Suggestions, Created Date.
- Authenticate with Google Sheets OAuth2.
- Specify suggestions spreadsheet and sheet.
14. **Add Sticky Notes for Documentation**
- Add Sticky Note nodes with provided descriptive content for:
- Workflow overview
- Jotform input
- Q&A and email reply
- Suggestions handling
- Comments logging
15. **Test Workflow**
- Submit test entries with each feedback type.
- Verify branching, AI processing, alerts, logging, and emails.
- Monitor for errors and resolve credential or API issues.
---
### 5. General Notes & Resources
| Note Content | Context or Link |
|-----------------------------------------------------------------------------------------------|-----------------------------------------------------------------------------------------------------|
| Workflow automates triage of Jotform feedback into categories with AI-powered processing. | General project description. |
| Urgent/angry messages trigger immediate Telegram alerts to meet a 6-hour SLA. | Support alerting SLA note. |
| Q&A Agent uses Google Sheets as a knowledge base to ground AI-generated answers. | See Sticky Note2 for detailed explanation. |
| Suggestions are summarized, alerted via Telegram, and logged for product team review. | See Sticky Note3 for workflow details. |
| Comments are logged with sentiment data for dashboards and periodic analysis. | See Sticky Note4 for purpose and usage. |
| Credentials (Jotform API, Google Palm API, Google Sheets OAuth2, Gmail OAuth2, Telegram API) | Must be configured securely in node credentials, not hardcoded. |
| Recommended to rename nodes to reflect your environment and maintain credentials separately. | Setup tip from Sticky Note. |
| For more info on LangChain nodes and Google Gemini integration, consult official n8n docs. | n8n official docs and Google Palm API references. |
---
**Disclaimer:** The provided text is exclusively derived from an automated workflow created with n8n, respecting all applicable content policies and containing only legal and public data.