Azure AI Search

Vector Database integration · 10 node(s).

00Overview

Manage Azure AI Search from a flow — create and delete indexes, upload, fetch and remove documents, and check index statistics and document counts before a bulk load. Run keyword, vector or hybrid queries against an index and optionally re-rank the results with a semantic configuration, so a flow can answer questions over your own content. Wire a query embedding straight from an AI Embed Text step into a vector search to build retrieval-augmented flows.

Every field below is exactly what you see in the Flomation editor. Fields marked ● live picker let you choose from a list pulled live from your account — no IDs to look up.

01Connecting Azure AI Search

In the Azure portal, open your search service under Search services and note its name — the {name} in https://{name}.search.windows.net. This goes in the node's Service Name field (or use Custom Endpoint with the full https://….search.windows.net URL for sovereign clouds or private endpoints).
Azure AI Search uses key-based authentication. Open Settings → Keys on the service and copy one of the admin keys (Primary or Secondary) for any node that creates or deletes indexes or writes documents; a query key is enough for search and read-only nodes.
Leave API Version blank to use the current GA version, or set it only when you need a preview feature or a specific sovereign-cloud version.
In Flomation, store the key as an environment secret (e.g. azureaisearch_secret), then pick it in each node's API Key field.

Field	Type		Details
API Key	`secret`	Required	Admin key from Settings ▸ Keys — a query key only covers reads
Custom Endpoint	`string`		https://my-search-service.search.windows.net — overrides Service Name (sovereign clouds, private endpoints)

Pick an Environment on your flow (Flow Settings → Environment) so the secret resolves. Secret fields never show the value — they reference ${secrets.your_secret}.

02Document

Azure AI Search: Count Documents

vectordatabase/azureaisearch/document_count · Action

Count the documents in an index. The count is eventually consistent — a just-finished upload can take a few seconds to show.

Field	Type		Details
Service Name	`string`		my-search-service — the {name} in https://{name}.search.windows.net
API Version	`string`		2024-07-01 (default)
Index Name	`string`	Required	products

Returns: count, result, tool_result, success, error

Azure AI Search: Delete Documents

vectordatabase/azureaisearch/document_delete · Action

Delete documents from an index by key. Deleting a key that does not exist counts as success (the service treats it as already gone). Needs an admin API key.

Field	Type		Details
Service Name	`string`		my-search-service — the {name} in https://{name}.search.windows.net
API Version	`string`		2024-07-01 (default)
Index Name	`string`	Required	products
Key Field	`string`	Required	id — the index's key field name
Document Keys	`string`	Required	doc-1, doc-2 — comma-separated keys to delete

Returns: results, count, tool_result, success, error

Azure AI Search: Get Document

vectordatabase/azureaisearch/document_get · Action

Look up a single document in an index by its key. Optionally select specific fields to return.

Field	Type		Details
Service Name	`string`		my-search-service — the {name} in https://{name}.search.windows.net
API Version	`string`		2024-07-01 (default)
Index Name	`string`	Required	products
Document Key	`string`	Required	doc-1
Select Fields	`string`		id,content — comma-separated fields to return (blank for all retrievable fields)

Returns: id, result, tool_result, success, error

Azure AI Search: Upload Documents

vectordatabase/azureaisearch/document_upload · Action

Add or update documents in an index. Merge or Upload (the default) upserts; Merge updates existing documents only; Upload replaces whole documents. Needs an admin API key.

Field	Type		Details
Service Name	`string`		my-search-service — the {name} in https://{name}.search.windows.net
API Version	`string`		2024-07-01 (default)
Index Name	`string`	Required	products
Documents (JSON array)	`object`	Required	[{"id":"1","content":"…","content_vector":[0.1,0.2]}] — each must carry the index's key field
Write Behaviour	`string`		choices: Merge or Upload (upsert), Upload (replace whole document), Merge (update existing only)

Returns: results, count, tool_result, success, error

03Index

Azure AI Search: Create or Update Index

vectordatabase/azureaisearch/index_create · Action

Create an Azure AI Search index from a full index definition (fields, vector search profiles, semantic configuration), or update it in place. Needs an admin API key.

Field	Type		Details
Service Name	`string`		my-search-service — the {name} in https://{name}.search.windows.net
API Version	`string`		2024-07-01 (default)
Index Name	`string`	Required	products — lowercase letters, digits and dashes
Index Definition (JSON)	`object`	Required	{"fields":[{"name":"id","type":"Edm.String","key":true},{"name":"content","type":"Edm.String","searchable":true},{"name":"content_vector","type":"Collection(Edm.Single)","dimensions":1536,"vectorSearchProfile":"default"}],"vectorSearch":{...}}
Only Create If Missing	`boolean`		Fail softly instead of overwriting when the index already exists

Returns: id, result, tool_result, success, error

Azure AI Search: Delete Index

vectordatabase/azureaisearch/index_delete · Action

Delete an Azure AI Search index and every document in it. This cannot be undone. Needs an admin API key.

Field	Type		Details
Service Name	`string`		my-search-service — the {name} in https://{name}.search.windows.net
API Version	`string`		2024-07-01 (default)
Index Name	`string`	Required	products

Returns: id, result, tool_result, success, error

Azure AI Search: Get Index

vectordatabase/azureaisearch/index_get · Action

Fetch an Azure AI Search index definition — fields, vector search profiles, scoring profiles, and semantic configuration.

Field	Type		Details
Service Name	`string`		my-search-service — the {name} in https://{name}.search.windows.net
API Version	`string`		2024-07-01 (default)
Index Name	`string`	Required	products

Returns: id, result, tool_result, success, error

Azure AI Search: Get Many Indexes

vectordatabase/azureaisearch/index_get_all · Action

List every index on the search service. Optionally select specific properties (e.g. name) to keep the output small.

Field	Type	Details
Service Name	`string`	my-search-service — the {name} in https://{name}.search.windows.net
API Version	`string`	2024-07-01 (default)
Select Properties	`string`	name — comma-separated index properties to return (blank for full definitions)

Returns: results, count, tool_result, success, error

Azure AI Search: Get Index Statistics

vectordatabase/azureaisearch/index_stats · Action

Fetch an index's document count and storage usage (including vector index size) — handy for capacity checks before a bulk upload.

Field	Type		Details
Service Name	`string`		my-search-service — the {name} in https://{name}.search.windows.net
API Version	`string`		2024-07-01 (default)
Index Name	`string`	Required	products

Returns: id, result, document_count, storage_size, tool_result, success, error

04Search

Azure AI Search: Search

vectordatabase/azureaisearch/search · Action

Query an index: keyword (full-text), vector (similarity over an embedding), or hybrid (both fused). Optionally re-rank with a semantic configuration. Results include @search.score.

Field	Type		Details
Service Name	`string`		my-search-service — the {name} in https://{name}.search.windows.net
API Version	`string`		2024-07-01 (default)
Index Name	`string`	Required	products
Search Text	`string`		What to look for — blank matches everything (*)
Search Mode	`string`		choices: Keyword (full-text), Vector (similarity), Hybrid (keyword + vector)
Query Vector (JSON array)	`text`		[0.12, -0.03, …] — the query embedding; wire in AI ▸ Embed Text
Vector Field	`string`		content_vector (default)
Nearest Neighbours (k)	`integer`		10 (default) — how many nearest documents the vector query considers
Filter (OData)	`string`		category eq 'technology' and rating gt 3
Select Fields	`string`		id,title,content — comma-separated fields to return
Order By	`string`		rating desc — ignored when relevance ranking applies
Max Results	`integer`		50 (default, max 1000)
Semantic Configuration	`string`		Name of a semantic configuration on the index — enables semantic re-ranking (Basic tier or higher)

Returns: results, count, tool_result, success, error

05Notes & Limitations

Behaviours and constraints worth knowing before you build with these nodes.

A query key is enough for the read nodes — running searches, fetching a document, reading or listing index definitions, and pulling counts and statistics — but creating or deleting an index and uploading or deleting documents are rejected unless you supply an admin key.
A document upload is reported as a failed step if any single document in the batch is rejected, and the result tells you how many failed and lists the offending keys even when the service accepted the rest of the batch.
Semantic re-ranking only applies when the named semantic configuration already exists in the index definition and the search service is on the Basic tier or higher; otherwise the query returns an error.
A single search returns at most 1,000 results, so narrow large result sets with a filter rather than expecting to page beyond that ceiling.