workflow-templates.md

layout	default
title	Workflow templates
nav_order	25

Workflow templates

OpenSearch provides several workflow templates for some common machine learning (ML) use cases. Using a template simplifies complex setups and provides many default values for use cases like semantic or conversational search.

You can specify a workflow template when you call the Create Workflow API:

• To use an OpenSearch-provided workflow template, specify the template use case as the use_case query parameter (see the Example). For a list of OpenSearch-provided templates, see Supported workflow templates.

• To use a custom workflow template, provide the complete template in the request body. For an example of a custom template, see an example JSON template or an example YAML template.

To provision the workflow, specify provision=true as a query parameter.

Example

In this example, you'll configure the semantic_search_with_cohere_embedding_query_enricher workflow template. The workflow created using this template performs the following configuration steps:

• Deploys an externally hosted Cohere model
• Creates an ingest pipeline using the model
• Creates a sample vector index and configures a search pipeline to define the default model ID for that index

Step 1: Create and provision the workflow

Send the following request to create and provision a workflow using the semantic_search_with_cohere_embedding_query_enricher workflow template. The only required request body field for this template is the API key for the Cohere Embed model:

POST /_plugins/_flow_framework/workflow?use_case=semantic_search_with_cohere_embedding_query_enricher&provision=true
{
    "create_connector.credential.key" : "<YOUR API KEY>"
}

{% include copy-curl.html %}

OpenSearch responds with a workflow ID for the created workflow:

{
  "workflow_id" : "8xL8bowB8y25Tqfenm50"
}

The workflow in the previous step creates a default vector index. The default index name is my-nlp-index:

{
  "create_index.name": "my-nlp-index"
}

For all default parameter values for this workflow template, see Cohere Embed semantic search defaults.

Step 2: Ingest documents into the index

To ingest documents into the index created in the previous step, send the following request:

PUT /my-nlp-index/_doc/1
{
  "passage_text": "Hello world",
  "id": "s1"
}

{% include copy-curl.html %}

Step 3: Perform vector search

To perform a vector search on your index, use a neural query clause:

GET /my-nlp-index/_search
{
  "_source": {
    "excludes": [
      "passage_embedding"
    ]
  },
  "query": {
    "neural": {
      "passage_embedding": {
        "query_text": "Hi world",
        "k": 100
      }
    }
  }
}

{% include copy-curl.html %}

Parameters

Each workflow template has a defined schema and a set of APIs with predefined default values for each step. For more information about template parameter defaults, see Supported workflow templates.

Overriding default values

To override a template's default values, provide the new values in the request body when sending a create workflow request. For example, the following request changes the Cohere model, the name of the text_embedding processor output field, and the name of the sparse index of the semantic_search_with_cohere_embedding template:

POST /_plugins/_flow_framework/workflow?use_case=semantic_search_with_cohere_embedding
{
    "create_connector.model" : "embed-multilingual-v3.0",
    "text_embedding.field_map.output": "book_embedding",
    "create_index.name": "sparse-book-index"
}

{% include copy-curl.html %}

Viewing workflow resources

The workflow you created provisioned all the necessary resources for semantic search. To view the provisioned resources, call the Get Workflow Status API and provide the workflowID for your workflow:

GET /_plugins/_flow_framework/workflow/8xL8bowB8y25Tqfenm50/_status

{% include copy-curl.html %}

Supported workflow templates

To use a workflow template, specify it in the use_case query parameter when creating a workflow. The following templates are supported:

The following templates are supported:

• Model deployment templates:
  • Amazon Bedrock Titan embedding
  • Amazon Bedrock Titan multimodal
  • Cohere embedding
  • Cohere chat
  • OpenAI embedding
  • OpenAI chat
• Semantic search templates:
  • Semantic search
  • Semantic search with a query enricher
  • Semantic search using a local model
  • Semantic search using a Cohere embedding model
  • Semantic search using Cohere embedding models with a query enricher
  • Semantic search using Cohere embedding models with reindexing
• Neural sparse search templates:
  • Neural sparse search
• Multimodal search templates:
  • Multimodal search
  • Multimodal search using Amazon Bedrock Titan
• Hybrid search templates:
  • Hybrid search
  • Hybrid search using a local model
• Conversational search templates:
  • Conversational search using an LLM
• Agentic search templates:
  • Agentic search with a flow agent
  • Agentic search with a conversational agent

Model deployment templates

The following workflow templates configure model deployment.

Amazon Bedrock Titan embedding

This workflow creates and deploys an Amazon Bedrock embedding model (by default, titan-embed-text-v1).

• Use case: bedrock_titan_embedding_model_deploy
• Created components: A connector and model for the Amazon Bedrock Titan embeddings model
• Required parameters:
  • create_connector.credential.access_key
  • create_connector.credential.secret_key
  • create_connector.credential.session_token
• Defaults

Note: Requires AWS credentials and access to Amazon Bedrock.

Amazon Bedrock Titan multimodal

This workflow creates and deploys an Amazon Bedrock multimodal embedding model (by default, titan-embed-image-v1).

• Use case: bedrock_titan_multimodal_model_deploy
• Created components: A connector and model for Amazon Bedrock Titan multimodal embeddings
• Required parameters:
  • create_connector.credential.access_key
  • create_connector.credential.secret_key
  • create_connector.credential.session_token
• Defaults

Note: Requires AWS credentials and access to Amazon Bedrock.

Cohere embedding

This workflow creates and deploys a Cohere embedding model (by default, embed-english-v3.0).

• Use case: cohere_embedding_model_deploy
• Created components: A connector and model for Cohere embedding
• Required parameters:
  • create_connector.credential.key
• Defaults

Note: Requires a Cohere API key.

Cohere chat

This workflow creates and deploys a Cohere chat model (by default, Cohere Command).

• Use case: cohere_chat_model_deploy
• Created components: A connector and model for Cohere chat
• Required parameters:
  • create_connector.credential.key
• Defaults

Note: Requires a Cohere API key.

OpenAI embedding

This workflow creates and deploys an OpenAI embedding model (by default, text-embedding-ada-002).

• Use case: open_ai_embedding_model_deploy
• Created components: A connector and model for OpenAI embeddings
• Required parameters:
  • create_connector.credential.key
• Defaults

Note: Requires an OpenAI API key.

OpenAI chat

This workflow creates and deploys an OpenAI chat model (by default, gpt-3.5-turbo).

• Use case: openai_chat_model_deploy
• Created components: A connector and model for OpenAI chat
• Required parameters:
  • create_connector.credential.key
• Defaults

Note: Requires an OpenAI API key.

Semantic search templates

The following workflow templates configure semantic search.

Semantic search

This workflow configures semantic search.

• Use case: semantic_search
• Created components:
  • An ingest pipeline with a text_embedding processor
  • A vector index configured with the pipeline
• Required parameters:
  • create_ingest_pipeline.model_id: The model ID of the text embedding model to be used
• Defaults

Semantic search with a query enricher

This workflow configures semantic search with a default model for neural queries.

• Use case: semantic_search_with_query_enricher
• Created components:
  • An ingest pipeline with a text_embedding processor
  • A vector index configured with the pipeline
  • A query_enricher search processor that sets a default model ID for neural queries.
• Required parameters:
  • create_ingest_pipeline.model_id: The model ID of the text embedding model to be used
• Defaults

Semantic search using a local model

This workflow configures semantic search and deploys a pretrained model.

• Use case: semantic_search_with_local_model
• Created components:
  • A pretrained model (by default, huggingface/sentence-transformers/paraphrase-MiniLM-L3-v2)
  • An ingest pipeline with a text_embedding processor
  • A vector index configured with the pipeline
  • A query_enricher search processor that sets a default model ID for neural queries.
• Required parameters: None
• Defaults

Note: Uses a local pretrained model with a default configuration.

Semantic search using a Cohere embedding model

This workflow configures semantic search and deploys a Cohere embedding model.

• Use case: semantic_search_with_cohere_embedding
• Created components:
  • A Cohere embedding model (by default, embed-english-v3.0) connector and deployment
  • An ingest pipeline with a text_embedding processor
  • A vector index configured with the pipeline
• Required parameters:
  • create_connector.credential.key: API key for the Cohere model
• Defaults

Note: Requires a Cohere API key.

Semantic search using Cohere embedding models with a query enricher

This workflow configures semantic search, deploys a Cohere embedding model, and adds a query enricher search processor.

• Use case: semantic_search_with_cohere_embedding_query_enricher
• Created components:
  • A Cohere embedding model connector and deployment
  • An ingest pipeline with a text_embedding processor
  • A vector index configured with the pipeline
  • A query_enricher search processor that sets a default model ID for neural queries.
• Required parameters:
  • create_connector.credential.key: API key for the Cohere model
• Defaults

Note: Requires a Cohere API key.

Semantic search using Cohere embedding models with reindexing

This workflow configures semantic search with a Cohere embedding model and reindexes an existing index.

• Use case: semantic_search_with_reindex
• Created components:
  • A Cohere embedding model connector and deployment
  • A vector index configured with the pipeline
  • A reindexing process
• Required parameters:
  • create_connector.credential.key: API key for the Cohere model
  • reindex.source_index: The source index to be reindexed
• Defaults

Note: Reindexes a source index into a newly configured k-NN index using a Cohere embedding model.

Neural sparse search templates

The following workflow template configures neural sparse search.

Neural sparse search

This workflow configures neural sparse search.

• Use case: local_neural_sparse_search_bi_encoder
• Created components:
  • A locally hosted pretrained sparse encoding model (by default, amazon/neural-sparse/opensearch-neural-sparse-encoding-v1)
  • An ingest pipeline with a sparse_encoding processor
  • A vector index configured with the pipeline
• Required parameters: None
• Defaults

Multimodal search templates

The following workflow templates configure multimodal search.

Multimodal search

This workflow configures multimodal search.

• Use case: multimodal_search
• Created components:
  • An ingest pipeline with a text_image_embedding processor
  • A vector index configured with the pipeline
• Required parameters:
  • create_ingest_pipeline.model_id: The model ID of the multimodal embedding model to be used
• Defaults

Multimodal search using Amazon Bedrock Titan

This workflow deploys an Amazon Bedrock multimodal model and configures a multimodal search pipeline.

• Use case: multimodal_search_with_bedrock_titan
• Created components:
  • An Amazon Bedrock Titan multimodal embedding model connector and deployment
  • An ingest pipeline with a text_image_embedding processor
  • A vector index for multimodal search configured with the pipeline
• Required parameters:
  • create_connector.credential.access_key
  • create_connector.credential.secret_key
  • create_connector.credential.session_token
• Defaults

Note: Requires AWS credentials and access to Amazon Bedrock.

Hybrid search templates

The following workflow templates configure hybrid search.

Hybrid search

This workflow configures hybrid search.

• Use case: hybrid_search
• Created components:
  • An ingest pipeline
  • A vector index configured with the pipeline
  • A search pipeline with a normalization_processor
• Required parameters:
  • create_ingest_pipeline.model_id: The model ID of the text embedding model to be used
• Defaults

Hybrid search using a local model

This workflow configures hybrid search and deploys a pretrained model.

• Use case: hybrid_search_with_local_model
• Created components:
  • A pretrained model (by default, huggingface/sentence-transformers/paraphrase-MiniLM-L3-v2)
  • An ingest pipeline
  • A vector index configured with the pipeline
  • A search pipeline with a normalization_processor
• Required parameters: None
• Defaults

Note: Uses a local pretrained model for hybrid search configuration.

Conversational search templates

The following workflow template configures conversational search with RAG.

Conversational search using an LLM

This workflow deploys a large language model and configures a conversational search pipeline.

• Use case: conversational_search_with_llm_deploy
• Created components:
  • A chat model (by default, Cohere Command) connector and deployment
  • A search pipeline with a retrieval_augmented_generation processor
• Required parameters:
  • create_connector.credential.key: API key for the LLM
• Defaults

Note: Requires an API key for the chosen language model.

Agentic search templates

The following workflow templates configure agentic search.

Agentic search with a flow agent

This workflow deploys an Amazon Bedrock chat model and configures an agentic search pipeline using a flow agent.

• Use case: agentic_search_with_flow_agent
• Created components:
  • An Amazon Bedrock connector and a remote chat model (by default, Claude 4 Sonnet)
  • A QueryPlanningTool
  • A flow agent wired to the QueryPlanningTool
  • A search pipeline with an agentic_query_translator request processor and an agentic_context response processor
• Required parameters:
  • create_connector.credential.access_key
  • create_connector.credential.secret_key
  • create_connector.credential.session_token
• Defaults

Note: Requires AWS credentials and access to Amazon Bedrock.

Agentic search with a conversational agent

This workflow deploys an Amazon Bedrock chat model and configures an agentic search pipeline using a conversational agent with conversation memory and multiple tools.

• Use case: agentic_search_with_conversational_agent
• Created components:
  • An Amazon Bedrock connector and a remote chat model (by default, Claude 4 Sonnet)
  • A QueryPlanningTool, ListIndexTool, and IndexMappingTool
  • A conversational agent with conversation memory and all three tools
  • A search pipeline with an agentic_query_translator request processor and an agentic_context response processor
• Required parameters:
  • create_connector.credential.access_key
  • create_connector.credential.secret_key
  • create_connector.credential.session_token
• Defaults

Note: Requires AWS credentials and access to Amazon Bedrock.