Automatisation Google Sheets avec n8n : résumé de feedbacks

{
  "id": "Lwvu2jjMU2irTyAY",
  "meta": {
    "instanceId": "fb924c73af8f703905bc09c9ee8076f48c17b596ed05b18c0ff86915ef8a7c4a"
  },
  "name": "Summarize Google Sheets form feedback via OpenAI's GPT-4",
  "tags": [
    {
      "id": "y9tvM3hISJKT2jeo",
      "name": "Ted's Tech Talks",
      "createdAt": "2023-08-15T22:12:34.260Z",
      "updatedAt": "2023-08-15T22:12:34.260Z"
    }
  ],
  "nodes": [
    {
      "id": "cd80cd2f-a6e1-48eb-ba05-0f8f1a0875e5",
      "name": "When clicking \"Test workflow\"",
      "type": "n8n-nodes-base.manualTrigger",
      "position": [
        680,
        320
      ],
      "parameters": {},
      "typeVersion": 1
    },
    {
      "id": "9f03f1c4-c47e-4eda-bc0a-a598c21e4616",
      "name": "Sticky Note",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        640,
        130
      ],
      "parameters": {
        "width": 369.1031874662338,
        "height": 349,
        "content": "### 1. Create a Google Sheet document\n* This tutorial uses Google Sheet document connected to Google Forms, but a standalone Sheet document will work too\n* Adapt initial trigger to your needs: run manually or at some time intervals\n\n[Link to the Google Sheets template](https://docs.google.com/spreadsheets/d/1Kcr1oF_RrfNQJczmJDpwClOSYpvSnwbeX-_pdUo91-I/edit?usp=sharing)"
      },
      "typeVersion": 1
    },
    {
      "id": "1e478f81-76e7-4fc3-a147-11a92d3f9998",
      "name": "Sticky Note1",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        1040,
        160
      ],
      "parameters": {
        "width": 394,
        "height": 319,
        "content": "### 2. Combine all answers into an array\n* Since the main goal is to provide an overall summary, we need to combine all answers for each Google Form question\n* Aggregate Node takes multiple incoming items and produces just a single item which contains arrays of user feedback"
      },
      "typeVersion": 1
    },
    {
      "id": "1ab06b51-3b9e-4a4c-afba-c98e529a636c",
      "name": "Sticky Note2",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        1480,
        160
      ],
      "parameters": {
        "width": 432,
        "height": 319,
        "content": "### 3. Generate a summary report\n* Enter a __system message__ with a overall instructions on how to analyze the feedback form\n* Provide a __user message__ with JSON arrays.\n\n__NB! Consider splitting the form questions for a very long forms or when the number of responses is too high__"
      },
      "typeVersion": 1
    },
    {
      "id": "ce0118a3-4eaf-4d60-adf0-5bde5d41328a",
      "name": "Sticky Note3",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        1940,
        160
      ],
      "parameters": {
        "width": 359.1031874662346,
        "height": 319,
        "content": "### 4. Convert to HTML and send an email\n* GPT is configured to reply in Markdown format. Markdown Node converts such text into HTML\n* Finally, the Gmail node sends a message with HTML report"
      },
      "typeVersion": 1
    },
    {
      "id": "37bc8ab5-328c-4f50-bbda-f7482bf36522",
      "name": "Get Google Sheets records",
      "type": "n8n-nodes-base.googleSheets",
      "position": [
        860,
        320
      ],
      "parameters": {
        "options": {},
        "sheetName": {
          "__rl": true,
          "mode": "list",
          "value": 2035968519,
          "cachedResultUrl": "https://docs.google.com/spreadsheets/d/1Kcr1oF_RrfNQJczmJDpwClOSYpvSnwbeX-_pdUo91-I/edit#gid=2035968519",
          "cachedResultName": "Form Responses 1"
        },
        "documentId": {
          "__rl": true,
          "mode": "list",
          "value": "1Kcr1oF_RrfNQJczmJDpwClOSYpvSnwbeX-_pdUo91-I",
          "cachedResultUrl": "https://docs.google.com/spreadsheets/d/1Kcr1oF_RrfNQJczmJDpwClOSYpvSnwbeX-_pdUo91-I/edit?usp=drivesdk",
          "cachedResultName": "Event feedback form (Responses)"
        }
      },
      "credentials": {
        "googleSheetsOAuth2Api": {
          "id": "RtRiRezoxiWkzZQt",
          "name": "Ted's Tech Talks Google account"
        }
      },
      "typeVersion": 4.2
    },
    {
      "id": "d75b11b1-2cce-40c2-ab5a-d18fdf7f5283",
      "name": "Aggregate responses into arrays",
      "type": "n8n-nodes-base.aggregate",
      "position": [
        1200,
        320
      ],
      "parameters": {
        "options": {},
        "fieldsToAggregate": {
          "fieldToAggregate": [
            {
              "fieldToAggregate": "['What went great?']"
            },
            {
              "fieldToAggregate": "['How can we improve?']"
            },
            {
              "fieldToAggregate": "['What is the chance of recommending our event?']"
            }
          ]
        }
      },
      "typeVersion": 1
    },
    {
      "id": "a90f83fe-809b-42db-b65d-43fb11b2979a",
      "name": "Summarize via GPT model",
      "type": "n8n-nodes-base.openAi",
      "position": [
        1620,
        320
      ],
      "parameters": {
        "prompt": {
          "messages": [
            {
              "role": "system",
              "content": "Your task is to summarize event feedback form responses. You will receive answers on three questions:\n1. What went great?\n2. How can we improve?\n3. What is the chance of recommending our event?\n\nEach questions has several answers separated by | character.\nAnalyze each question and prepare a summary report. It should contain an overall sentiment regarding the event, followed by the constructive ideas of what to improve.\n\nReply in Markdown formatting"
            },
            {
              "content": "=1. What went great: ```{{ $json['What went great?'].join(' | ') }}```\n2. How can we improve: ```{{ $json['How can we improve?'].join(' | ') }}```\n3. What is the chance of recommending our event: ```{{ $json['What is the chance of recommending our event?'].join(' | ') }}```"
            }
          ]
        },
        "options": {
          "temperature": 0.3
        },
        "resource": "chat",
        "chatModel": "gpt-4-turbo-preview"
      },
      "credentials": {
        "openAiApi": {
          "id": "rveqdSfp7pCRON1T",
          "name": "Ted's Tech Talks OpenAi"
        }
      },
      "typeVersion": 1.1
    },
    {
      "id": "2c8d4e46-9d3e-4655-952b-37d04f673914",
      "name": "Convet from Markdown to HTML",
      "type": "n8n-nodes-base.markdown",
      "position": [
        1980,
        320
      ],
      "parameters": {
        "mode": "markdownToHtml",
        "options": {
          "completeHTMLDocument": false
        },
        "markdown": "={{ $json.message.content }}"
      },
      "typeVersion": 1
    },
    {
      "id": "a27d8664-dc87-4458-9f12-970b88ab6515",
      "name": "Send via Gmail",
      "type": "n8n-nodes-base.gmail",
      "position": [
        2160,
        320
      ],
      "parameters": {
        "sendTo": "teds.tech.talks@gmail.com",
        "message": "={{ $json.data }}",
        "options": {
          "appendAttribution": false
        },
        "subject": "Feedback form response"
      },
      "credentials": {
        "gmailOAuth2": {
          "id": "UllrXlZsDnkdA3tT",
          "name": "Gmail account"
        }
      },
      "typeVersion": 2.1
    }
  ],
  "active": false,
  "pinData": {},
  "settings": {
    "executionOrder": "v1"
  },
  "versionId": "756cdd85-49dd-4f0f-acc7-58f834a3512f",
  "connections": {
    "Summarize via GPT model": {
      "main": [
        [
          {
            "node": "Convet from Markdown to HTML",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Get Google Sheets records": {
      "main": [
        [
          {
            "node": "Aggregate responses into arrays",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Convet from Markdown to HTML": {
      "main": [
        [
          {
            "node": "Send via Gmail",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "When clicking \"Test workflow\"": {
      "main": [
        [
          {
            "node": "Get Google Sheets records",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Aggregate responses into arrays": {
      "main": [
        [
          {
            "node": "Summarize via GPT model",
            "type": "main",
            "index": 0
          }
        ]
      ]
    }
  }
}