Jenny This workflow corresponds to n8n.io template #7945 — we link there as the canonical source.
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": "FnlDCNDV3x4pYVyC",
"meta": {
"templateId": "self-building-ai-agent",
"templateCredsSetupCompleted": true
},
"name": "Hybrid Search with Qdrant & n8n, Legal AI: Indexing",
"tags": [],
"nodes": [
{
"id": "2556a724-93f9-4ecc-8112-10458fea8b3e",
"name": "Create Collection",
"type": "n8n-nodes-qdrant.qdrant",
"position": [
560,
368
],
"parameters": {
"vectors": "{\n \"mxbai_large\": \n {\n \"size\": 1024,\n \"distance\": \"Cosine\"\n }\n}",
"operation": "createCollection",
"shardNumber": {},
"sparseVectors": "{\n \"bm25\": \n {\n \"modifier\": \"idf\"\n }\n}",
"collectionName": "legalQA_test",
"requestOptions": {},
"replicationFactor": {},
"writeConsistencyFactor": {}
},
"credentials": {
"qdrantApi": {
"name": "<your credential>"
}
},
"typeVersion": 1
},
{
"id": "c4c7120a-aff6-4bdd-880b-903761b88af8",
"name": "Check Collection Exists",
"type": "n8n-nodes-qdrant.qdrant",
"position": [
208,
288
],
"parameters": {
"operation": "collectionExists",
"collectionName": "legalQA_test",
"requestOptions": {}
},
"credentials": {
"qdrantApi": {
"name": "<your credential>"
}
},
"typeVersion": 1
},
{
"id": "0639e81c-130c-4fd0-a4df-80509c2f0aaf",
"name": "If",
"type": "n8n-nodes-base.if",
"position": [
400,
288
],
"parameters": {
"options": {},
"conditions": {
"options": {
"version": 2,
"leftValue": "",
"caseSensitive": true,
"typeValidation": "loose"
},
"combinator": "and",
"conditions": [
{
"id": "d67b3ed7-aea5-4307-86f0-76c06a9da5fa",
"operator": {
"name": "filter.operator.equals",
"type": "string",
"operation": "equals"
},
"leftValue": "={{ $json.result.exists }}",
"rightValue": "true"
}
]
},
"looseTypeValidation": true
},
"typeVersion": 2.2
},
{
"id": "c454200a-9216-4e69-88cf-bcb3f93b65f0",
"name": "Sticky Note",
"type": "n8n-nodes-base.stickyNote",
"position": [
-1056,
192
],
"parameters": {
"width": 592,
"height": 864,
"content": "## Index Legal Dataset to Qdrant for Hybrid Retrieval\n*This pipeline is the first part of **\"Hybrid Search with Qdrant & n8n, Legal AI\"**. \nThe second part, **\"Hybrid Search with Qdrant & n8n, Legal AI: Retrieval\"**, covers retrieval and simple evaluation.* \n\n### Overview\nThis pipeline transforms a [Q&A legal corpus from Hugging Face (isaacus)](https://huggingface.co/datasets/isaacus/LegalQAEval) into vector representations and indexes them to Qdrant, providing the foundation for running [**Hybrid Search**](https://qdrant.tech/articles/hybrid-search/), combining:\n\n- [**Dense vectors**](https://qdrant.tech/documentation/concepts/vectors/#dense-vectors) (embeddings) for semantic similarity search; \n- [**Sparse vectors**](https://qdrant.tech/documentation/concepts/vectors/#sparse-vectors) for keyword-based exact search.\n\n\nAfter running this pipeline, you will have a Qdrant collection with your legal dataset ready for hybrid retrieval on [BM25](https://en.wikipedia.org/wiki/Okapi_BM25) and dense embeddings: either [mxbai-embed-large-v1](https://huggingface.co/mixedbread-ai/mxbai-embed-large-v1) or [`text-embedding-3-small`](https://platform.openai.com/docs/models/text-embedding-3-small).\n\n#### Options for Embedding Inference\nThis pipeline equips you with two approaches for generating dense vectors:\n\n1. Using [**Qdrant Cloud Inference**](https://qdrant.tech/documentation/cloud/inference/), conversion to vectors handled directly in Qdrant;\n2. Using external provider, e.g. OpenAI for generating embeddings.\n\n#### Prerequisites\n- A cluster on [Qdrant Cloud](https://cloud.qdrant.io/) \n - Paid cluster in the US region if you want to use **Qdrant Cloud Inference** \n - Free Tier Cluster if using an external provider (here OpenAI) \n- Qdrant Cluster credentials: \n - You'll be guided on how to obtain both the **URL** and **API_KEY** from the Qdrant Cloud UI when setting up your cluster; \n- An **OpenAI API key** (if you\u2019re not using Qdrant\u2019s Cloud Inference); \n\n#### P.S.\n- To ask retrieval in Qdrant-related questions, join the [Qdrant Discord](https://discord.gg/ArVgNHV6). \n- Star [Qdrant n8n community node repo](https://github.com/qdrant/n8n-nodes-qdrant) <3"
},
"typeVersion": 1
},
{
"id": "03b3d5c1-cbed-43c6-8d2a-241c8a04d79d",
"name": "Index Dataset from HuggingFace",
"type": "n8n-nodes-base.manualTrigger",
"position": [
-368,
768
],
"parameters": {},
"typeVersion": 1
},
{
"id": "8e97d7e3-1daf-4cb8-89ea-6235b0d5f8ad",
"name": "Split Them All Out",
"type": "n8n-nodes-base.splitOut",
"position": [
256,
944
],
"parameters": {
"options": {},
"fieldToSplitOut": "splits"
},
"typeVersion": 1
},
{
"id": "4e9a2449-ef56-4f76-b6b6-9195a591e2a8",
"name": "Get Dataset Splits",
"type": "n8n-nodes-base.httpRequest",
"position": [
64,
944
],
"parameters": {
"url": "https://datasets-server.huggingface.co/splits",
"options": {},
"sendQuery": true,
"queryParameters": {
"parameters": [
{
"name": "dataset",
"value": "={{ $json.dataset }}"
}
]
}
},
"typeVersion": 4.2
},
{
"id": "4227306b-4008-4d3a-a233-404d12729114",
"name": "Divide Per Row",
"type": "n8n-nodes-base.splitOut",
"position": [
640,
944
],
"parameters": {
"options": {},
"fieldToSplitOut": "rows"
},
"typeVersion": 1
},
{
"id": "8d9b6c80-00ff-48c5-a9aa-75318c10e080",
"name": "Loop Over Batches",
"type": "n8n-nodes-base.splitInBatches",
"position": [
2640,
496
],
"parameters": {
"options": {
"reset": false
},
"batchSize": 8
},
"executeOnce": false,
"typeVersion": 3
},
{
"id": "987ee18a-78b8-46f4-be12-5897176784e0",
"name": "Aggregate a Batch",
"type": "n8n-nodes-base.aggregate",
"position": [
2976,
512
],
"parameters": {
"options": {},
"aggregate": "aggregateAllItemData",
"destinationFieldName": "batch"
},
"typeVersion": 1
},
{
"id": "5a11322c-665d-41e4-86fa-b7a0b16a4c75",
"name": "Upsert Points",
"type": "n8n-nodes-qdrant.qdrant",
"position": [
3232,
512
],
"parameters": {
"points": "=[\n {{\n $json.batch.map(i => \n ({ \n \"id\": i.idx,\n \"payload\": { \n \"text\": i.text, \n \"ids_qa\": i.ids_qa\n },\n \"vector\": {\n \"mxbai_large\": {\n \"text\": i.text,\n \"model\": \"mixedbread-ai/mxbai-embed-large-v1\"\n },\n \"bm25\": {\n \"text\": i.text,\n \"model\": \"qdrant/bm25\",\n \"options\": {\n \"avg_len\": i.avg_len\n }\n }\n }\n }).toJsonString()\n )\n }}\n]",
"resource": "point",
"operation": "upsertPoints",
"collectionName": {
"__rl": true,
"mode": "list",
"value": "legalQA_test",
"cachedResultName": "legalQA_test"
},
"requestOptions": {}
},
"credentials": {
"qdrantApi": {
"name": "<your credential>"
}
},
"typeVersion": 1
},
{
"id": "a4d4ed4a-b24a-4dba-895c-46964d2915be",
"name": "Limit",
"type": "n8n-nodes-base.limit",
"position": [
1440,
1264
],
"parameters": {
"maxItems": 500
},
"typeVersion": 1
},
{
"id": "3d45c4b2-c3da-4add-9256-a9cdba062637",
"name": "Merge",
"type": "n8n-nodes-base.merge",
"position": [
2224,
784
],
"parameters": {
"mode": "combine",
"options": {},
"combineBy": "combineAll"
},
"typeVersion": 3.2
},
{
"id": "8a5ba479-f1b1-4bdf-8934-ff39dfa384dd",
"name": "Sum them Up",
"type": "n8n-nodes-base.summarize",
"position": [
1856,
1264
],
"parameters": {
"options": {},
"fieldsToSummarize": {
"values": [
{
"field": "words_in_text",
"aggregation": "sum"
}
]
}
},
"typeVersion": 1.1
},
{
"id": "dced86c8-5dfb-4718-89ce-707997268382",
"name": "Get the Average Text Length",
"type": "n8n-nodes-base.set",
"position": [
2064,
1264
],
"parameters": {
"options": {},
"assignments": {
"assignments": [
{
"id": "0f436085-17d6-4131-8e6d-7ffee50b60be",
"name": "avg_len",
"type": "number",
"value": "={{ $json.sum_words_in_text / 500 }}"
}
]
}
},
"typeVersion": 3.4
},
{
"id": "c6de3504-36f4-47b9-8a1d-7df398284e8e",
"name": "Loop Over Batches1",
"type": "n8n-nodes-base.splitInBatches",
"position": [
2640,
1312
],
"parameters": {
"options": {
"reset": false
},
"batchSize": 8
},
"executeOnce": false,
"typeVersion": 3
},
{
"id": "19e6b91d-f03a-4cb7-afd9-a148eb724877",
"name": "Upsert Points1",
"type": "n8n-nodes-qdrant.qdrant",
"position": [
4192,
1312
],
"parameters": {
"points": "=[\n {{\n $json.batch.map(i => \n ({ \n \"id\": i.idx,\n \"payload\": { \n \"text\": i.text, \n \"ids_qa\": i.ids_qa\n },\n \"vector\": {\n \"open_ai_small\": i.embedding,\n \"bm25\": {\n \"text\": i.text,\n \"model\": \"qdrant/bm25\",\n \"options\": {\n \"avg_len\": i.avg_len\n }\n }\n }\n }).toJsonString()\n )\n }}\n]",
"resource": "point",
"operation": "upsertPoints",
"collectionName": {
"__rl": true,
"mode": "list",
"value": "legalQA_openAI_test",
"cachedResultName": "legalQA_openAI_test"
},
"requestOptions": {}
},
"credentials": {
"qdrantApi": {
"name": "<your credential>"
}
},
"typeVersion": 1
},
{
"id": "1b4ceeb5-fa40-4544-a4f8-cfd9860de452",
"name": "Create Collection1",
"type": "n8n-nodes-qdrant.qdrant",
"position": [
3008,
1840
],
"parameters": {
"vectors": "{\n \"open_ai_small\": \n {\n \"size\": 1536,\n \"distance\": \"Cosine\"\n }\n}",
"operation": "createCollection",
"shardNumber": {},
"sparseVectors": "{\n \"bm25\": \n {\n \"modifier\": \"idf\"\n }\n}",
"collectionName": "legalQA_openAI_test",
"requestOptions": {},
"replicationFactor": {},
"writeConsistencyFactor": {}
},
"credentials": {
"qdrantApi": {
"name": "<your credential>"
}
},
"typeVersion": 1
},
{
"id": "948b1d9a-a529-4919-bb99-63ce30e2e2a5",
"name": "Check Collection Exists1",
"type": "n8n-nodes-qdrant.qdrant",
"position": [
2608,
1744
],
"parameters": {
"operation": "collectionExists",
"collectionName": "legalQA_openAI_test",
"requestOptions": {}
},
"credentials": {
"qdrantApi": {
"name": "<your credential>"
}
},
"typeVersion": 1
},
{
"id": "e73d6246-e782-4293-bd57-ccd9a9276e06",
"name": "If1",
"type": "n8n-nodes-base.if",
"position": [
2816,
1744
],
"parameters": {
"options": {},
"conditions": {
"options": {
"version": 2,
"leftValue": "",
"caseSensitive": true,
"typeValidation": "loose"
},
"combinator": "and",
"conditions": [
{
"id": "d67b3ed7-aea5-4307-86f0-76c06a9da5fa",
"operator": {
"name": "filter.operator.equals",
"type": "string",
"operation": "equals"
},
"leftValue": "={{ $json.result.exists }}",
"rightValue": "true"
}
]
},
"looseTypeValidation": true
},
"typeVersion": 2.2
},
{
"id": "7809aff3-02d1-45e4-949d-b251b37be7ef",
"name": "Merge1",
"type": "n8n-nodes-base.merge",
"position": [
3680,
1312
],
"parameters": {
"mode": "combine",
"options": {},
"combineBy": "combineByPosition"
},
"typeVersion": 3.2
},
{
"id": "d68cf8a5-400f-41e3-b8bf-3a3e71ff1985",
"name": "Split Out",
"type": "n8n-nodes-base.splitOut",
"position": [
3520,
1104
],
"parameters": {
"options": {},
"fieldToSplitOut": "data"
},
"typeVersion": 1
},
{
"id": "cdac0c35-6aa9-441a-9859-3f3bfa8e3521",
"name": "Get OpenAI embeddings",
"type": "n8n-nodes-base.httpRequest",
"position": [
3344,
1104
],
"parameters": {
"url": "https://api.openai.com/v1/embeddings",
"method": "POST",
"options": {},
"sendBody": true,
"authentication": "predefinedCredentialType",
"bodyParameters": {
"parameters": [
{
"name": "input",
"value": "={{ $json.batch.map(item => item.text) }}"
},
{
"name": "model",
"value": "text-embedding-3-small"
}
]
},
"nodeCredentialType": "openAiApi"
},
"credentials": {
"openAiApi": {
"name": "<your credential>"
}
},
"typeVersion": 4.2
},
{
"id": "3a5ba038-021f-4cfc-8d59-189357309479",
"name": "Sticky Note1",
"type": "n8n-nodes-base.stickyNote",
"position": [
0,
592
],
"parameters": {
"color": 5,
"width": 1344,
"height": 528,
"content": "## Get Dataset from Hugging Face\n\nFetching a sample dataset from Hugging Face using the [Dataset Viewer API](https://huggingface.co/docs/dataset-viewer/quick_start).\n**Dataset:** [LegalQAEval from isaacus](https://huggingface.co/datasets/isaacus/LegalQAEval).\n\n1. **Retrieve dataset splits**. \n2. **Fetch all items with pagination** \n - Apply [pagination in HTTP node](https://docs.n8n.io/code/cookbook/http-node/pagination/#enable-pagination) to retrieve the full dataset. \n3. **Deduplicate text chunks** \n - The dataset contains duplicate `text` chunks, since multiple questions may belong to each passage. \n - Deduplicate before indexing into Qdrant to avoid storing duplicates. \n - Aggregate the corresponding **question\u2013answer IDs** so they can be reused later during retrieval evaluation. \n4. **Format data for batching** (embeddings inference & indexing to Qdrant) \n"
},
"typeVersion": 1
},
{
"id": "4f9d02bb-6474-4448-9eab-5bc599cc2587",
"name": "Get Dataset Rows (Pagination)",
"type": "n8n-nodes-base.httpRequest",
"position": [
448,
944
],
"parameters": {
"url": "=https://datasets-server.huggingface.co/rows",
"options": {
"pagination": {
"pagination": {
"parameters": {
"parameters": [
{
"name": "offset",
"value": "={{ $pageCount * 100 }}"
}
]
},
"requestInterval": 1000,
"completeExpression": "={{ $pageCount * 100 > $response.body.num_rows_total}}\n",
"paginationCompleteWhen": "other"
}
}
},
"sendQuery": true,
"queryParameters": {
"parameters": [
{
"name": "dataset",
"value": "={{ $json.dataset }}"
},
{
"name": "config",
"value": "={{ $json.config }}"
},
{
"name": "split",
"value": "={{ $json.split }}"
},
{
"name": "length",
"value": "=100"
}
]
}
},
"typeVersion": 4.2
},
{
"id": "d1b63d11-d424-44ca-8ca9-843eb488235a",
"name": "Sticky Note2",
"type": "n8n-nodes-base.stickyNote",
"position": [
1424,
1024
],
"parameters": {
"color": 5,
"width": 800,
"height": 416,
"content": "## Estimate Average Length of Text Chunks\n\nAverage length of texts in the dataset is a part of the [BM25](https://en.wikipedia.org/wiki/Okapi_BM25) formula used for keyword-based retrieval.\n\n1. **Select a subsample** \n2. **Count words per text chunk** \n3. **Compute average length** \n - Calculate the mean across all chunks in the subsample. \n - This value will be used as the **average document length (avg_len)** parameter in BM25."
},
"typeVersion": 1
},
{
"id": "b16cbdd6-789c-4b21-8755-502e089ca547",
"name": "Sticky Note3",
"type": "n8n-nodes-base.stickyNote",
"position": [
16,
-128
],
"parameters": {
"color": 5,
"width": 1088,
"height": 640,
"content": "## Create [Qdrant Collection](https://qdrant.tech/documentation/concepts/collections/) for Hybrid Search\nThe collection used for **Hybrid Search** is configured here with two types of vectors:\n\n**1. [Dense Vectors](https://qdrant.tech/documentation/concepts/vectors/#dense-vectors)**\nIn this pipeline, we're using the [**mxbai-embed-large-v1**](https://huggingface.co/mixedbread-ai/mxbai-embed-large-v1) embedding model through Qdrant's Cloud Inference. Hence, we need to specify during the collection configuration its:\n- **Dimensions**: 1024 \n- **Similarity metric**: `cosine`\n\n\n**2. [Sparse Vectors](https://qdrant.tech/documentation/concepts/vectors/#sparse-vectors)**\nQdrant\u2019s main mechanism for setting up **keyword-based retrieval**. \nFor example, you can set up retrieval with:\n - [**BM25**](https://en.wikipedia.org/wiki/Okapi_BM25) (used in this pipeline);\n - Qdrant provides an [**`IDF` modifier**](https://qdrant.tech/documentation/concepts/indexing/#idf-modifier) for sparse vectors. This enables Qdrant to calculate **inverse document frequency (IDF)** statistics on the server side. These statistics evaluate the importance of keywords, for example, in BM25. \n - SPLADE, miniCOIL and other sparse neural retrievers. \n\n"
},
"typeVersion": 1
},
{
"id": "3f4cedea-edeb-4796-967b-d75b95fd4aad",
"name": "Sticky Note4",
"type": "n8n-nodes-base.stickyNote",
"position": [
2544,
288
],
"parameters": {
"color": 5,
"width": 960,
"height": 480,
"content": "## (Option \u21161) Index Text Chunks to Qdrant Using [Cloud Inference](https://qdrant.tech/documentation/cloud/inference/)\n\n- **Embed & upsert text chunks in batches** \n - **Dense embeddings inference + upsert handled by Qdrant node**, it takes care of generating embeddings and inserting them into the collection. \n - **Sparse representations for BM25** are created automatically under the hood by Qdrant. \n"
},
"typeVersion": 1
},
{
"id": "67fc6b7c-9168-4214-94cd-3c2d68e477cc",
"name": "Sticky Note5",
"type": "n8n-nodes-base.stickyNote",
"position": [
2528,
1552
],
"parameters": {
"color": 7,
"width": 688,
"height": 448,
"content": "## (Option \u21162) 1. Configure a Collection for OpenAI Embeddings & BM25 Retrieval\nSince [`text-embedding-3-small`] OpenAI embeddings have a different dimensionality (1536) than mxbai embeddings (1024), you need to account for this when configuring the collection. \n \nFor simplicity, create a **separate collection** dedicated to OpenAI embeddings. This collection will be used to index texts in this block. "
},
"typeVersion": 1
},
{
"id": "ed76cf94-3b3b-4c8f-af1f-2ea5f7096785",
"name": "Sticky Note6",
"type": "n8n-nodes-base.stickyNote",
"position": [
2512,
864
],
"parameters": {
"color": 5,
"width": 1872,
"height": 1152,
"content": "## (Option \u21162) Index Text Chunks to Qdrant Using External Embedding Provider (OpenAI)\n*Don't forget to create and configure a separate collection for OpenAI\u2019s [`text-embedding-3-small`](https://platform.openai.com/docs/models/text-embedding-3-small) embeddings.*\n\n1. **Embed texts in batches** with OpenAI's [`text-embedding-3-small`](https://platform.openai.com/docs/models/text-embedding-3-small), generating dense vectors. \n\n2. **Upsert batches to Qdrant:**\n- Pass pre-embedded by OpenAi dense vectors to Qdrant;\n- Sparse representations for BM25 are created automatically under the hood by Qdrant. "
},
"typeVersion": 1
},
{
"id": "5eb0cbf7-a151-4bf4-a180-914909a04901",
"name": "Restructure for Deduplicating",
"type": "n8n-nodes-base.set",
"position": [
816,
944
],
"parameters": {
"options": {},
"assignments": {
"assignments": [
{
"id": "961c95d9-c803-404b-b4b6-cb66a8a33928",
"name": "id_qa",
"type": "string",
"value": "={{ $json.row.id }}"
},
{
"id": "00f4a104-8515-49fe-a094-89d22a2ead05",
"name": "text",
"type": "string",
"value": "={{ $json.row.text }}"
}
]
}
},
"typeVersion": 3.4
},
{
"id": "e3f582f9-aad1-47a4-83a8-1e0127b78ce9",
"name": "Restructure for Batching",
"type": "n8n-nodes-base.set",
"position": [
1200,
944
],
"parameters": {
"options": {},
"assignments": {
"assignments": [
{
"id": "23528728-83f3-4f11-9d66-feddc3bf27d1",
"name": "idx",
"type": "number",
"value": "={{ $itemIndex }}"
},
{
"id": "f663bae7-ff0c-440f-9a57-cb363322fc9c",
"name": "text",
"type": "string",
"value": "={{ $json.text }}"
},
{
"id": "bfb956b4-d5e2-46b2-b41a-850a4e00765f",
"name": "ids_qa",
"type": "array",
"value": "={{ $json.appended_id_qa }}"
}
]
}
},
"typeVersion": 3.4
},
{
"id": "74568439-a6ab-4f4e-acc5-9a0784d6c1d2",
"name": "Deduplicate Texts",
"type": "n8n-nodes-base.summarize",
"position": [
1008,
944
],
"parameters": {
"options": {},
"fieldsToSplitBy": "text",
"fieldsToSummarize": {
"values": [
{
"field": "id_qa",
"aggregation": "append"
}
]
}
},
"typeVersion": 1.1
},
{
"id": "b65a9c60-44e1-465c-99f4-1d33428e5c4a",
"name": "Calculate #words in Each Text",
"type": "n8n-nodes-base.set",
"position": [
1648,
1264
],
"parameters": {
"options": {},
"assignments": {
"assignments": [
{
"id": "29dc2299-fb1e-4b0a-bff1-0a3e88f7eb03",
"name": "words_in_text",
"type": "number",
"value": "={{ $json.text.trim().split(/\\s+/).length }}"
}
]
}
},
"typeVersion": 3.4
},
{
"id": "f778e469-8a74-47fe-a854-7da473156f87",
"name": "Edit Fields",
"type": "n8n-nodes-base.set",
"position": [
2912,
1104
],
"parameters": {
"options": {}
},
"typeVersion": 3.4
},
{
"id": "5a66c3c1-2c6b-4280-b7cb-514f2ae5c720",
"name": "Aggregate a Batch to Embed",
"type": "n8n-nodes-base.aggregate",
"position": [
3088,
1216
],
"parameters": {
"options": {},
"aggregate": "aggregateAllItemData",
"destinationFieldName": "batch"
},
"typeVersion": 1
},
{
"id": "1e4971c7-c41f-4e7b-b9a1-c777193578c7",
"name": "Aggregate a Batch to Upsert",
"type": "n8n-nodes-base.aggregate",
"position": [
3952,
1312
],
"parameters": {
"options": {},
"aggregate": "aggregateAllItemData",
"destinationFieldName": "batch"
},
"typeVersion": 1
}
],
"active": false,
"settings": {
"executionOrder": "v1"
},
"versionId": "fc4f19dc-4bac-4a41-944d-2c3d0b469e33",
"connections": {
"If": {
"main": [
[],
[
{
"node": "Create Collection",
"type": "main",
"index": 0
}
]
]
},
"If1": {
"main": [
[],
[
{
"node": "Create Collection1",
"type": "main",
"index": 0
}
]
]
},
"Limit": {
"main": [
[
{
"node": "Calculate #words in Each Text",
"type": "main",
"index": 0
}
]
]
},
"Merge": {
"main": [
[
{
"node": "Loop Over Batches",
"type": "main",
"index": 0
},
{
"node": "Loop Over Batches1",
"type": "main",
"index": 0
}
]
]
},
"Merge1": {
"main": [
[
{
"node": "Aggregate a Batch to Upsert",
"type": "main",
"index": 0
}
]
]
},
"Split Out": {
"main": [
[
{
"node": "Merge1",
"type": "main",
"index": 0
}
]
]
},
"Sum them Up": {
"main": [
[
{
"node": "Get the Average Text Length",
"type": "main",
"index": 0
}
]
]
},
"Upsert Points": {
"main": [
[
{
"node": "Loop Over Batches",
"type": "main",
"index": 0
}
]
]
},
"Divide Per Row": {
"main": [
[
{
"node": "Restructure for Deduplicating",
"type": "main",
"index": 0
}
]
]
},
"Upsert Points1": {
"main": [
[
{
"node": "Loop Over Batches1",
"type": "main",
"index": 0
}
]
]
},
"Aggregate a Batch": {
"main": [
[
{
"node": "Upsert Points",
"type": "main",
"index": 0
}
]
]
},
"Create Collection": {
"main": [
[]
]
},
"Deduplicate Texts": {
"main": [
[
{
"node": "Restructure for Batching",
"type": "main",
"index": 0
}
]
]
},
"Loop Over Batches": {
"main": [
[],
[
{
"node": "Aggregate a Batch",
"type": "main",
"index": 0
}
]
]
},
"Get Dataset Splits": {
"main": [
[
{
"node": "Split Them All Out",
"type": "main",
"index": 0
}
]
]
},
"Loop Over Batches1": {
"main": [
[
{
"node": "Edit Fields",
"type": "main",
"index": 0
}
],
[
{
"node": "Merge1",
"type": "main",
"index": 1
},
{
"node": "Aggregate a Batch to Embed",
"type": "main",
"index": 0
}
]
]
},
"Split Them All Out": {
"main": [
[
{
"node": "Get Dataset Rows (Pagination)",
"type": "main",
"index": 0
}
]
]
},
"Get OpenAI embeddings": {
"main": [
[
{
"node": "Split Out",
"type": "main",
"index": 0
}
]
]
},
"Check Collection Exists": {
"main": [
[
{
"node": "If",
"type": "main",
"index": 0
}
]
]
},
"Check Collection Exists1": {
"main": [
[
{
"node": "If1",
"type": "main",
"index": 0
}
]
]
},
"Restructure for Batching": {
"main": [
[
{
"node": "Limit",
"type": "main",
"index": 0
},
{
"node": "Merge",
"type": "main",
"index": 0
}
]
]
},
"Aggregate a Batch to Embed": {
"main": [
[
{
"node": "Get OpenAI embeddings",
"type": "main",
"index": 0
}
]
]
},
"Aggregate a Batch to Upsert": {
"main": [
[
{
"node": "Upsert Points1",
"type": "main",
"index": 0
}
]
]
},
"Get the Average Text Length": {
"main": [
[
{
"node": "Merge",
"type": "main",
"index": 1
}
]
]
},
"Calculate #words in Each Text": {
"main": [
[
{
"node": "Sum them Up",
"type": "main",
"index": 0
}
]
]
},
"Get Dataset Rows (Pagination)": {
"main": [
[
{
"node": "Divide Per Row",
"type": "main",
"index": 0
}
]
]
},
"Restructure for Deduplicating": {
"main": [
[
{
"node": "Deduplicate Texts",
"type": "main",
"index": 0
}
]
]
},
"Index Dataset from HuggingFace": {
"main": [
[
{
"node": "Get Dataset Splits",
"type": "main",
"index": 0
},
{
"node": "Check Collection Exists",
"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.
openAiApiqdrantApi
Pro
For the full experience including quality scoring and batch install features for each workflow upgrade to Pro
About this workflow
This pipeline is the first part of "Hybrid Search with Qdrant & n8n, Legal AI"*. The second part, "Hybrid Search with Qdrant & n8n, Legal AI: Retrieval", covers retrieval and simple evaluation.*
Source: https://n8n.io/workflows/7945/ — original creator credit. Request a take-down →
Related workflows
Workflows that share integrations, category, or trigger type with this one. All free to copy and import.
This is the second part of "Hybrid Search with Qdrant & n8n, Legal AI."* The first part, "Indexing", covers preparing and uploading the dataset to Qdrant.*
HTTP Request, N8N Nodes Qdrant
The benefits being (1) the vision model doesn't need to keep all document scans in context (expensive) and (2) ability to query on graphical content such as charts, graphs and tables. Page extracts fr
HTTP Request, N8N Nodes Qdrant, Chat Trigger +7
Featherless.ai is an inference provider with a different pricing model - they charge a flat subscription fee (starting from $10) and allows for unlimited token usage instead. If you're typically spend
N8N Nodes Featherless, HTTP Request, N8N Nodes Qdrant +6
Self-Hosted
Google Drive Trigger, Google Drive, N8N Nodes Qdrant +4
[1/3 - anomaly detection] [1/2 - KNN classification] Batch upload dataset to Qdrant (crops dataset). Uses manualTrigger, googleCloudStorage, httpRequest, stickyNote. Event-driven trigger; 25 nodes.
Google Cloud Storage, HTTP Request