Workflow: Manual Openai Export

Workflow Details

Download Workflow
{
    "meta": {
        "instanceId": "cb484ba7b742928a2048bf8829668bed5b5ad9787579adea888f05980292a4a7"
    },
    "nodes": [
        {
            "id": "d9bae984-2ce7-4f6b-ab53-527ac9dfea3d",
            "name": "When clicking \"Execute Workflow\"",
            "type": "n8n-nodes-base.manualTrigger",
            "position": [
                680,
                320
            ],
            "parameters": [],
            "typeVersion": 1
        },
        {
            "id": "32ecf73c-b6e9-4bd6-9ecc-d82c4c50d7b5",
            "name": "Reddit",
            "type": "n8n-nodes-base.reddit",
            "position": [
                880,
                320
            ],
            "parameters": {
                "keyword": "n8n",
                "location": "allReddit",
                "operation": "search",
                "additionalFields": {
                    "sort": "new"
                }
            },
            "credentials": [],
            "typeVersion": 1
        },
        {
            "id": "4b560620-a101-4566-b066-4ce3f44d8b0c",
            "name": "Sticky Note1",
            "type": "n8n-nodes-base.stickyNote",
            "position": [
                120,
                180
            ],
            "parameters": {
                "width": 507.1052631578949103641207329928874969482421875,
                "height": 210.99380804953551660219090990722179412841796875,
                "content": "## What this workflow does\n\u2714\ufe0e 1) Get posts from reddit that might be about n8n\n    - Filter for the most relevant posts (posted in last 7 days and more than 5 upvotes and is original content)\n\n\u2714\ufe0e 2) Check if the post is actually about n8n\n\n\u2714\ufe0e 3) if it is, categorise with OpenAi.\n"
            },
            "typeVersion": 1
        },
        {
            "id": "f3be9af5-b4ff-4f4e-a726-fc05fab94521",
            "name": "Set",
            "type": "n8n-nodes-base.set",
            "position": [
                1260,
                320
            ],
            "parameters": {
                "values": {
                    "number": [
                        {
                            "name": "upvotes",
                            "value": "={{ $json.ups }}"
                        },
                        {
                            "name": "subredditSize",
                            "value": "={{ $json.subreddit_subscribers }}"
                        }
                    ],
                    "string": [
                        {
                            "name": "selftextTrimmed",
                            "value": "={{ $json.selftext.substring(0,500) }}"
                        },
                        {
                            "name": "subreddit",
                            "value": "={{ $json.subreddit }}"
                        },
                        {
                            "name": "date",
                            "value": "={{ DateTime.fromSeconds($json.created).toLocaleString() }}"
                        },
                        {
                            "name": "url",
                            "value": "={{ $json.url }}"
                        }
                    ]
                },
                "options": [],
                "keepOnlySet": true
            },
            "typeVersion": 1
        },
        {
            "id": "b1dbf78f-c7c6-4ab7-a957-78d58c5e13e3",
            "name": "IF",
            "type": "n8n-nodes-base.if",
            "position": [
                1060,
                320
            ],
            "parameters": {
                "conditions": {
                    "number": [
                        {
                            "value1": "={{ $json.ups }}",
                            "value2": "=5",
                            "operation": "largerEqual"
                        }
                    ],
                    "string": [
                        {
                            "value1": "={{ $json.selftext }}",
                            "operation": "isNotEmpty"
                        }
                    ],
                    "dateTime": [
                        {
                            "value1": "={{ DateTime.fromSeconds($json.created).toISO() }}",
                            "value2": "={{ $today.minus({days: 7}).toISO() }}"
                        }
                    ]
                }
            },
            "typeVersion": 1
        },
        {
            "id": "a3aa9e43-a824-4cc1-b4e6-d41a2e8e56cd",
            "name": "Sticky Note2",
            "type": "n8n-nodes-base.stickyNote",
            "position": [
                120,
                660
            ],
            "parameters": {
                "width": 504.4736842105266987346112728118896484375,
                "height": 116.7797420572506581493144039995968341827392578125,
                "content": "## Drawbacks\n\ud83e\udd14 Workflow only considers first 500 characters of each reddit post. So if n8n is mentioned after this amount, it won't register as being a post about n8n.io."
            },
            "typeVersion": 1
        },
        {
            "id": "b3d566aa-1645-4c2c-9704-15aa2e42bb12",
            "name": "IF1",
            "type": "n8n-nodes-base.if",
            "position": [
                1880,
                340
            ],
            "parameters": {
                "conditions": {
                    "string": [
                        {
                            "value1": "={{ $json.choices[0].text }}",
                            "value2": "No",
                            "operation": "contains"
                        }
                    ]
                }
            },
            "typeVersion": 1
        },
        {
            "id": "0ad54272-08b9-46d4-8e6a-1fb55a92d3e4",
            "name": "Merge",
            "type": "n8n-nodes-base.merge",
            "position": [
                1680,
                520
            ],
            "parameters": {
                "mode": "combine",
                "options": {
                    "fuzzyCompare": false,
                    "includeUnpaired": true
                },
                "combinationMode": "mergeByPosition"
            },
            "typeVersion": 2
        },
        {
            "id": "288f53cc-0e53-4683-ac0e-debe0a3691b8",
            "name": "Merge1",
            "type": "n8n-nodes-base.merge",
            "position": [
                2340,
                540
            ],
            "parameters": {
                "mode": "combine",
                "options": {
                    "fuzzyCompare": false,
                    "includeUnpaired": true
                },
                "combinationMode": "mergeByPosition"
            },
            "typeVersion": 2
        },
        {
            "id": "46280db5-e4b0-4108-958a-763b6410caa0",
            "name": "SetFinal",
            "type": "n8n-nodes-base.set",
            "position": [
                2560,
                540
            ],
            "parameters": {
                "values": {
                    "number": [
                        {
                            "name": "upvotes",
                            "value": "={{ $json.upvotes }}"
                        },
                        {
                            "name": "subredditSize",
                            "value": "={{ $json.subredditSize }}"
                        }
                    ],
                    "string": [
                        {
                            "name": "subreddit",
                            "value": "={{ $json.subreddit }}"
                        },
                        {
                            "name": "bulletSummary",
                            "value": "={{ $json.text }}"
                        },
                        {
                            "name": "date",
                            "value": "={{ $json.date }}"
                        },
                        {
                            "name": "url",
                            "value": "={{ $json.url }}"
                        }
                    ]
                },
                "options": [],
                "keepOnlySet": true
            },
            "typeVersion": 1
        },
        {
            "id": "ac8c4847-4d73-4dce-9543-a199e8b11b51",
            "name": "Sticky Note3",
            "type": "n8n-nodes-base.stickyNote",
            "position": [
                120,
                400
            ],
            "parameters": {
                "width": 507.1052631578949103641207329928874969482421875,
                "height": 247.53869969040255227810121141374111175537109375,
                "content": "## Next steps\n* Improve OpenAI Summary node prompt to return cleaner summaries.\n* Extend to **more platforms\/sources** - e.g. it would be really cool to monitor larger slack communities in this way. \n* Do some classification on type of user to highlight users likely to be in our **ICP**.\n* Separate a list of data sources (reddit, twitter, slack, discord etc.), extract messages from there and have them go to a **sub workflow for classification and summarisation.**"
            },
            "typeVersion": 1
        },
        {
            "id": "12ab5ba4-d24d-4fa1-a0d1-d1e81e2d5dee",
            "name": "OpenAI Summary",
            "type": "n8n-nodes-base.openAi",
            "notes": "A one sentence summary of what the post is about.",
            "disabled": true,
            "position": [
                2160,
                160
            ],
            "parameters": {
                "input": "={{ $json.selftextTrimmed }}",
                "options": {
                    "temperature": 0.299999999999999988897769753748434595763683319091796875
                },
                "operation": "edit",
                "instruction": "Summarise what this is talking about in a meta way less than 20 words. Ignore punctuation in your summary and return a short, human readable summary."
            },
            "credentials": [],
            "typeVersion": 1
        },
        {
            "id": "e303a1aa-ee93-4f8f-b834-19aa8da7fe95",
            "name": "OpenAI Classify",
            "type": "n8n-nodes-base.openAi",
            "notes": "Is the post about n8n?",
            "position": [
                1460,
                320
            ],
            "parameters": {
                "prompt": "=Decide whether a reddit post is about n8n.io, a workflow automation low code tool that can be self-hosted, or not.\nReddit Post:  {{ $json.selftextTrimmed }}\nAbout n8n?: Yes\/No",
                "options": {
                    "maxTokens": 32
                },
                "simplifyOutput": false
            },
            "credentials": [],
            "notesInFlow": true,
            "typeVersion": 1
        },
        {
            "id": "f56cb8b6-4c28-448e-b259-8946ffc4c1f7",
            "name": "OpenAI Summary Backup",
            "type": "n8n-nodes-base.openAi",
            "notes": "A one sentence summary of what the post is about.",
            "position": [
                2160,
                340
            ],
            "parameters": {
                "prompt": "=Summarise what this is talking about in a meta way in only 1 sentence.\n\n {{ $json.selftextTrimmed }}",
                "options": {
                    "maxTokens": 128
                }
            },
            "credentials": [],
            "typeVersion": 1
        },
        {
            "id": "d1eacbf2-9cc8-482d-a7d2-34c351f20871",
            "name": "Sticky Note",
            "type": "n8n-nodes-base.stickyNote",
            "position": [
                640,
                520
            ],
            "parameters": {
                "width": 843.4114964984020161864464171230792999267578125,
                "height": 258.6767901193690022410009987652301788330078125,
                "content": "## What we learned\n- \ud83e\udeb6 **Writing prompts**: small changes in the type of prompt result in very different results. e.g. for Summarising OpenAI would use multiple sentences even if we asked it to use only 1. We got better results by following OpenAI's documentation.\n    - We could make OpenAI node easier to work with for new users by the node inputs being oriented not to sending parameters to api but by following [their suggestions](https:\/\/platform.openai.com\/docs\/guides\/completion\/prompt-design) - e.g. have a field for expected output format rather than just for prompt.\n- \u2195\ufe0f **Changing the max_tokens parameter** drastically changes results - sometimes making it smaller even improves results (e.g. when you want a yes\/no response in the OpenAI Classify node). In their [docs](https:\/\/platform.openai.com\/docs\/guides\/completion\/inserting-text) they recommend using max_tokens>256 but [n8n by default](https:\/\/community.n8n.io\/t\/openai-result-not-complete\/21533) uses max_tokens=16. We should probably update this."
            },
            "typeVersion": 1
        }
    ],
    "connections": {
        "IF": {
            "main": [
                [
                    {
                        "node": "Set",
                        "type": "main",
                        "index": 0
                    }
                ]
            ]
        },
        "IF1": {
            "main": [
                null,
                [
                    {
                        "node": "OpenAI Summary Backup",
                        "type": "main",
                        "index": 0
                    },
                    {
                        "node": "Merge1",
                        "type": "main",
                        "index": 1
                    }
                ]
            ]
        },
        "Set": {
            "main": [
                [
                    {
                        "node": "OpenAI Classify",
                        "type": "main",
                        "index": 0
                    },
                    {
                        "node": "Merge",
                        "type": "main",
                        "index": 1
                    }
                ]
            ]
        },
        "Merge": {
            "main": [
                [
                    {
                        "node": "IF1",
                        "type": "main",
                        "index": 0
                    }
                ]
            ]
        },
        "Merge1": {
            "main": [
                [
                    {
                        "node": "SetFinal",
                        "type": "main",
                        "index": 0
                    }
                ]
            ]
        },
        "Reddit": {
            "main": [
                [
                    {
                        "node": "IF",
                        "type": "main",
                        "index": 0
                    }
                ]
            ]
        },
        "OpenAI Classify": {
            "main": [
                [
                    {
                        "node": "Merge",
                        "type": "main",
                        "index": 0
                    }
                ]
            ]
        },
        "OpenAI Summary Backup": {
            "main": [
                [
                    {
                        "node": "Merge1",
                        "type": "main",
                        "index": 0
                    }
                ]
            ]
        },
        "When clicking \"Execute Workflow\"": {
            "main": [
                [
                    {
                        "node": "Reddit",
                        "type": "main",
                        "index": 0
                    }
                ]
            ]
        }
    }
}
Back to Workflows

Related Workflows

HTTP Stickynote Automate Webhook
View
Respondtowebhook Stickynote Automate Webhook
View
Splitout Manual Automation Webhook
View
🌐 Confluence Page AI Powered Chatbot
View
Bitrix24 Open Chanel RAG Chatbot Application Workflow example with Webhook Integration
View
Googleslides Slack Automate Triggered
View
HR-focused automation pipeline with AI
View
Optimize Prompt
View
Hunter Pipedrive Create Triggered
View
Save new Files received on Telegram to Google Drive
View