Extract Company Data From Websites with Gemini AI Through Chat Conversation

The workflow JSON

Copy or download the full n8n JSON below. Paste it into a new n8n workflow, add your credentials, activate. Full import guide →

{
  "id": "UY2rhMDjLVtGntWm",
  "meta": {
    "templateCredsSetupCompleted": true
  },
  "name": "AI-Driven Company Data Extraction and Localization",
  "tags": [],
  "nodes": [
    {
      "id": "0d59c94f-1e64-41f2-b6b5-da88dfb2d1b1",
      "name": "When chat message received",
      "type": "@n8n/n8n-nodes-langchain.chatTrigger",
      "position": [
        -752,
        0
      ],
      "parameters": {
        "public": true,
        "options": {
          "responseMode": "responseNodes"
        },
        "initialMessages": "Hi! I can help you extract company information. Please enter the URL to get started."
      },
      "typeVersion": 1.4
    },
    {
      "id": "5b143ed0-e661-4ab1-8695-cf8c7ee50237",
      "name": "Google Gemini Chat Model",
      "type": "@n8n/n8n-nodes-langchain.lmChatGoogleGemini",
      "position": [
        -288,
        208
      ],
      "parameters": {
        "options": {}
      },
      "credentials": {
        "googlePalmApi": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 1
    },
    {
      "id": "ca977928-4980-4ffb-94d1-5f16911c6ce7",
      "name": "ExtractedChatInput",
      "type": "n8n-nodes-base.code",
      "position": [
        64,
        0
      ],
      "parameters": {
        "jsCode": "const rawText = $input.first().json.output || ''\nlet jsonString = rawText\n\nconst codeBlockRegex = /```(?:json)?\\s*([\\s\\S]*?)\\s*```/i\nconst match = rawText.match(codeBlockRegex)\n\nif (match && match[1]) {\n  jsonString = match[1]\n}\n\nlet resultObject = { url: null }\n\ntry {\n  resultObject = JSON.parse(jsonString)\n} catch (e) {\n}\n\nreturn [\n  {\n    json: resultObject\n  }\n]"
      },
      "typeVersion": 2
    },
    {
      "id": "8132e3b0-9fe4-4372-8bf0-8aa35b7de734",
      "name": "Request User Input",
      "type": "@n8n/n8n-nodes-langchain.chat",
      "position": [
        480,
        16
      ],
      "parameters": {
        "message": "Please enter the company website URL you want to search.",
        "options": {}
      },
      "typeVersion": 1
    },
    {
      "id": "a69cf0cc-3cd1-46f7-978b-0e4c5e3053a0",
      "name": "Google Gemini Chat Model1",
      "type": "@n8n/n8n-nodes-langchain.lmChatGoogleGemini",
      "position": [
        688,
        176
      ],
      "parameters": {
        "options": {}
      },
      "credentials": {
        "googlePalmApi": {
          "name": "<your credential>"
        }
      },
      "typeVersion": 1
    },
    {
      "id": "78c24fcb-46b5-481a-af76-79e4f058e19c",
      "name": "Config",
      "type": "n8n-nodes-base.set",
      "position": [
        -528,
        0
      ],
      "parameters": {
        "options": {},
        "assignments": {
          "assignments": [
            {
              "id": "3a6a2908-f5cd-4b8d-98d1-c3260b5ddfa5",
              "name": "chatInput",
              "type": "string",
              "value": "={{ $json.chatInput }}"
            },
            {
              "id": "12378c7e-817a-4d34-9c6f-e1d4c16579f1",
              "name": "targetCompanyFields",
              "type": "string",
              "value": "CEO,Address,Summary"
            },
            {
              "id": "90186d6e-7cf3-425e-94e5-b409a5a56161",
              "name": "language",
              "type": "string",
              "value": "English"
            }
          ]
        }
      },
      "typeVersion": 3.4
    },
    {
      "id": "057b79d3-9363-4623-832f-a03e69aa929b",
      "name": "HttpRequestTool",
      "type": "n8n-nodes-base.httpRequestTool",
      "position": [
        832,
        176
      ],
      "parameters": {
        "url": "={{ /*n8n-auto-generated-fromAI-override*/ $fromAI('URL', `URL to Get HTML`, 'string') }}",
        "options": {
          "timeout": 60000,
          "redirect": {
            "redirect": {
              "maxRedirects": 10
            }
          },
          "response": {
            "response": {
              "neverError": true,
              "fullResponse": true
            }
          },
          "allowUnauthorizedCerts": true
        },
        "sendHeaders": true,
        "toolDescription": "Get Page Content. Retrieves the content from the specified URL.",
        "headerParameters": {
          "parameters": [
            {
              "name": "=User-Agent",
              "value": "=Mozilla/5.0 (Macintosh; Intel Mac OS X 14_5_0) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/129.0.6668.71 Safari/537.36"
            }
          ]
        }
      },
      "typeVersion": 4.2
    },
    {
      "id": "866b2c6f-3fe5-4506-8ee4-eb56d7d4a048",
      "name": "ParsedCsv",
      "type": "n8n-nodes-base.code",
      "position": [
        1008,
        -16
      ],
      "parameters": {
        "jsCode": "const data = $input.first().json\n\nlet csvText = ''\nconst match = data.output.match(/```csv([\\s\\S]*?)```/)\n\nif (match) {\n  csvText = match[1].trim()\n}\n\nif (!csvText) {\n  throw new Error('CSV text not found')\n}\n\nconst lines = csvText.trim().split('\\n').map(line => line.replace(/\\r$/, ''))\n\nfunction parseCSVLine(line) {\n  const cells = []\n  let current = ''\n  let insideQuotes = false\n\n  for (let i = 0; i < line.length; i++) {\n    const char = line[i]\n    if (char === '\"') {\n      if (insideQuotes && line[i + 1] === '\"') {\n        current += '\"'\n        i++\n      } else {\n        insideQuotes = !insideQuotes\n      }\n    } else if (char === ',' && !insideQuotes) {\n      cells.push(current)\n      current = ''\n    } else {\n      current += char\n    }\n  }\n  cells.push(current)\n  return cells\n}\n\nconst headers = parseCSVLine(lines[0]).map(h => h.trim())\n\nconst records = lines.slice(1).map(line => {\n  if (!line.trim()) {\n    return null\n  }\n\n  const cells = parseCSVLine(line)\n  const obj = {}\n\n  headers.forEach((key, idx) => {\n    obj[key] = (cells[idx] || '').trim()\n  })\n  return obj\n}).filter(r => r !== null)\n\nreturn records"
      },
      "typeVersion": 2
    },
    {
      "id": "31f9c590-632e-4bfa-bee8-50d4673d36f1",
      "name": "Respond to Chat",
      "type": "@n8n/n8n-nodes-langchain.chat",
      "position": [
        1424,
        -16
      ],
      "parameters": {
        "message": "={{ $json.text }}",
        "options": {},
        "waitUserReply": false
      },
      "typeVersion": 1
    },
    {
      "id": "528fcf55-207b-482d-b766-9801a8e66513",
      "name": "ChatResponse",
      "type": "n8n-nodes-base.code",
      "position": [
        1216,
        -16
      ],
      "parameters": {
        "jsCode": "const IGNORED_KEYS = ['autoFetchStatus', 'autoFetchReason']\nconst data = $input.first().json\nconst reason = data.autoFetchReason || '(No reason provided)'\n\nif (data.autoFetchStatus !== 'ok') {\n  return {\n    json: {\n      text: \"\u26a0\ufe0f **Extraction finished with issues**\" + '\\n\\n' + reason\n    }\n  }\n}\n\nconst header = \"\u2705 **Successfully extracted company information**\"\nconst fields = Object.keys(data)\n  .filter(key => !IGNORED_KEYS.includes(key))\n  .map(key => {\n    const label = key\n    const value = data[key] || 'N/A'\n    return `\u2022 **${label}**\\n\\n> ${value}`\n  })\n  .join('\\n\\n')\n\nreturn {\n  json: {\n    text: header + '\\n\\n' +\n          reason + '\\n\\n' +\n          '\\n\\n' +\n          '---' + '\\n\\n' +\n          '\\n\\n' +\n          fields\n  }\n}"
      },
      "typeVersion": 2
    },
    {
      "id": "819c8bbe-fcbc-4a76-a15d-ccffe93d5a62",
      "name": "Sticky Note",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        -1472,
        -128
      ],
      "parameters": {
        "width": 672,
        "height": 512,
        "content": "## How it works\nPlease send a corporate website URL via chat.\n\nThe AI will investigate the company website on your behalf and return the extracted company information.  \nSince this is set up as a conversational workflow, retrying or trying another URL is simple.\n\n## How to use\n* To get started, please set up the Credential in the `Gemini` node attached to the `AI Agent` node.\n  * You can obtain an API key from [Google AI Studio](https://aistudio.google.com/api-keys).\n\nOnce configured, the workflow will run when you send a corporate website URL (e.g., https://example.com/) via chat.\n\n## Customizing this workflow\nYou can change the settings in the `Config` node.\n\n* You can modify `targetCompanyFields` to customize which company data fields are extracted.\n* You can modify `language` to receive the results in a language other than English."
      },
      "typeVersion": 1
    },
    {
      "id": "519ff98c-29fd-4b01-9081-d31b9adc706d",
      "name": "Request Next URL",
      "type": "@n8n/n8n-nodes-langchain.chat",
      "position": [
        1632,
        -16
      ],
      "parameters": {
        "message": "Would you like to search for another company? Please enter the URL.",
        "options": {}
      },
      "typeVersion": 1
    },
    {
      "id": "aa787064-7d59-4361-98be-05b606666184",
      "name": "AI Agent (Extract URL)",
      "type": "@n8n/n8n-nodes-langchain.agent",
      "position": [
        -288,
        0
      ],
      "parameters": {
        "text": "=Extract the homepage URL from the User Input and return it in the specified Output Format inside a code block.\n\nImportant:\nAlways return a valid JSON object.\nIf no URL is found, you MUST return the object with null: { \"url\": null }\n\n[Output Format]\n{ \"url\": \"extracted_url_or_null\" }\n\n[User Input]\n{{ $json.chatInput }}",
        "options": {},
        "promptType": "define"
      },
      "typeVersion": 3
    },
    {
      "id": "dfab6b48-f012-4b10-966e-a518f45bc050",
      "name": "Check URL",
      "type": "n8n-nodes-base.if",
      "position": [
        272,
        0
      ],
      "parameters": {
        "options": {},
        "conditions": {
          "options": {
            "version": 2,
            "leftValue": "",
            "caseSensitive": true,
            "typeValidation": "strict"
          },
          "combinator": "and",
          "conditions": [
            {
              "id": "07702149-ff99-438e-86d9-17dec54e21c3",
              "operator": {
                "type": "string",
                "operation": "notEmpty",
                "singleValue": true
              },
              "leftValue": "={{ $json.url }}",
              "rightValue": ""
            }
          ]
        }
      },
      "typeVersion": 2.2
    },
    {
      "id": "b99fb313-e065-4a60-850e-b2e7707d3d69",
      "name": "AI Agent (Access URL)",
      "type": "@n8n/n8n-nodes-langchain.agent",
      "onError": "continueRegularOutput",
      "position": [
        688,
        -16
      ],
      "parameters": {
        "text": "=Please extract the following information based on the System Instructions.\n\n[Target URL]\n{{ $json.url }}\n\n[Target Fields to Extract]\n{{ $('Config').first().json.targetCompanyFields }}\n\n[Translation Language]\n{{ $('Config').first().json.language }}\n\nGenerate the CSV response now.",
        "options": {
          "maxIterations": "=15",
          "systemMessage": "=You are an expert web scraper. Extract company info into strict CSV format.\n\n### Core Directive\n**Think Step-by-Step.**\nYour goal is to extract information **strictly from the Official Corporate Website**.\nSimulate a human researcher who validates the site's legitimacy before searching for data.\n\n### Tool Usage\n- Use `HttpRequestTool` to fetch pages.\n- **Iterative Workflow**: You MUST use the tool multiple times to follow links.\n\n### Execution Process\n1. **Analyze & Validate (CRITICAL)**:\n   - **Primary Rule**: The site MUST be the Official Corporate Website. **Anything else MUST result in an 'error' status.**\n   - **Mandatory Corporate Data Check**: The site MUST clearly confirm the **Official Company Name (Legal Entity)**. If the legal entity name cannot be confirmed on the landing page, Validation fails.\n   - **IMMEDIATE REJECTION CRITERIA**:\n     - **Non-Corporate Signal**: Any site that displays **\"Operator Information\"** instead of \"Company Profile\". This definitively means it is NOT a corporate website.\n     - **Media/Blogs**: Comparison sites, Rankings, Reviews, News articles.\n     - **Directories**: Yelp, Job boards, SNS.\n   - **Action upon Validation Failure**: If the site meets ANY rejection criteria, proceed directly to Step 4 (Status Logic) and set 'autoFetchStatus' to 'error'. **DO NOT continue to navigation (Step 2).**\n\n2. **Navigate to Company Profile (MANDATORY)**:\n   - **Prerequisite**: This step can only begin IF validation in Step 1 passed ('ok').\n   - **Exclusion**: You MUST NOT use Service pages (e.g., links labeled 'Products', 'Services', 'Pricing') as a source for core corporate data.\n   - **Assumption**: Detailed info (CEO, Address) is rarely on the homepage.\n   - **Action**: You MUST search the menu or footer for links dedicated to corporate structure: \"Company Profile\", \"About Us (Corporate)\", \"Access\", or \"Contact\".\n   - **Constraint**: Do NOT report \"N/A\" for core fields (like CEO, Address) without checking the specific \"Company Profile\" page.\n\n3. **Extraction**:\n   - Extract strictly from the domain. No hallucinations.\n   - Translate all extracted values into the language specified in the user request. Do NOT translate headers.\n\n4. **Status Logic and Output (CRITICAL)**:\n   - **Status Definition**: 'ok' if Validation passed AND core data (Legal Name, CEO, Address) was SUCCESSFULLY extracted from the Official Corporate Page (Step 2). 'error' otherwise.\n   - **Requirement**: `autoFetchReason` is MANDATORY. State clearly why the site was rejected (e.g., \"Found 'Operator Information', therefore not a corporate site\") or your successful navigation path.\n   - **Format**: Single Markdown code block (```csv ... ```). Row 1 MUST include requested fields plus \"autoFetchStatus\" and \"autoFetchReason\".\n   - **CSV SYNTAX (ABSOLUTE)**: All values in the data row MUST be enclosed in double quotes."
        },
        "promptType": "define"
      },
      "retryOnFail": true,
      "typeVersion": 3,
      "alwaysOutputData": true,
      "waitBetweenTries": 5000
    },
    {
      "id": "3ec14a4c-eb08-46dc-9cc2-41bdd0153aca",
      "name": "Sticky Note1",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        -784,
        -128
      ],
      "parameters": {
        "color": 7,
        "width": 1424,
        "height": 512,
        "content": "## Extract URLs using AI\nThis step uses AI to extract URLs from the chat message. If no URL is detected, it prompts the user to enter a URL again."
      },
      "typeVersion": 1
    },
    {
      "id": "9187913d-a41b-46ab-a207-d77ffcc30c6e",
      "name": "Sticky Note2",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        656,
        -128
      ],
      "parameters": {
        "color": 7,
        "width": 1184,
        "height": 512,
        "content": "## AI Search\nThe AI accesses the URL to extract company information and returns the results."
      },
      "typeVersion": 1
    }
  ],
  "active": false,
  "settings": {
    "executionOrder": "v1"
  },
  "versionId": "30227fcf-d248-4a8f-a428-e49a483b2a0f",
  "connections": {
    "Config": {
      "main": [
        [
          {
            "node": "AI Agent (Extract URL)",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Check URL": {
      "main": [
        [
          {
            "node": "AI Agent (Access URL)",
            "type": "main",
            "index": 0
          }
        ],
        [
          {
            "node": "Request User Input",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "ParsedCsv": {
      "main": [
        [
          {
            "node": "ChatResponse",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "ChatResponse": {
      "main": [
        [
          {
            "node": "Respond to Chat",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "HttpRequestTool": {
      "ai_tool": [
        [
          {
            "node": "AI Agent (Access URL)",
            "type": "ai_tool",
            "index": 0
          }
        ]
      ]
    },
    "Respond to Chat": {
      "main": [
        [
          {
            "node": "Request Next URL",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Request Next URL": {
      "main": [
        [
          {
            "node": "AI Agent (Extract URL)",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "ExtractedChatInput": {
      "main": [
        [
          {
            "node": "Check URL",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Request User Input": {
      "main": [
        [
          {
            "node": "AI Agent (Extract URL)",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "AI Agent (Access URL)": {
      "main": [
        [
          {
            "node": "ParsedCsv",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "AI Agent (Extract URL)": {
      "main": [
        [
          {
            "node": "ExtractedChatInput",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Google Gemini Chat Model": {
      "ai_languageModel": [
        [
          {
            "node": "AI Agent (Extract URL)",
            "type": "ai_languageModel",
            "index": 0
          }
        ]
      ]
    },
    "Google Gemini Chat Model1": {
      "ai_languageModel": [
        [
          {
            "node": "AI Agent (Access URL)",
            "type": "ai_languageModel",
            "index": 0
          }
        ]
      ]
    },
    "When chat message received": {
      "main": [
        [
          {
            "node": "Config",
            "type": "main",
            "index": 0
          }
        ]
      ]
    }
  }
}

Credentials you'll need

Each integration node will prompt for credentials when you import. We strip credential IDs before publishing — you'll add your own.

• googlePalmApi

About this workflow

Please send a corporate website URL via chat.

Source: https://n8n.io/workflows/11682/ — original creator credit. Request a take-down →

More AI & RAG workflows → · Browse all categories →

Related workflows

Workflows that share integrations, category, or trigger type with this one. All free to copy and import.