Resources · n8n integration

n8n Weaviate Vector Store nodeConfigure Weaviate Vector Store in n8n.

Weaviate is where a workflow keeps its searchable memory. The n8n Weaviate Vector Store node writes documents into a Weaviate collection and reads them back by meaning instead of by keyword. Two settings drive it, and four operation modes cover insert, search and agent use.

Try the Weaviate Vector Store node on n8nSee the 2 parameters

Verified Trustpilot reviews · AI, automation & growth agency

Why automate

What does the n8n Weaviate Vector Store node actually do?

It connects a workflow to a Weaviate Cluster and treats one collection as a vector store, a database that indexes text by its embedding, the numeric fingerprint of its meaning. A search then returns what is close in sense, not what matches a word. The node runs like any other action node, once per incoming item, and its Operation Mode decides whether that run puts documents in or pulls them out.

Say a support team keeps its procedures in scattered documents. A workflow loads them, the Default Data Loader attaches the metadata, and the node runs in Insert Documents mode against the chosen Weaviate Collection. Embedding Batch Size controls how many documents get embedded in one batch, 200 by default, which matters when a first import pushes thousands of pages at once.

Search is the mirror image. In Get Many mode, Prompt carries the question, Limit sets how many ranked results come back, and Include Metadata decides whether the source fields travel with the text. Rerank Results reorders what came back before the workflow uses it. The output goes straight into an email draft, a ticket reply or a report, without any model in the loop.

The third pattern hands the collection to a reasoning loop. Retrieve Documents (As Tool for AI Agent) turns the node into a tool that an AI Agent calls on its own, which is why the Name and Description fields exist: the description tells the model what the store contains and when to look inside it. The same logic applies when the node feeds an AI Agent Tool setup rather than a single agent.

Picking between modes is the real decision. Retrieve Documents (As Vector Store for Chain/Tool) is the one to use with a Vector Store Retriever feeding a Question and Answer Chain, when the retrieval should be silent and systematic. The tool mode is better when the agent should decide for itself whether a lookup is needed. The Vector Store Question Answer Tool suits cases where the answer must be summarized rather than returned raw. Whatever model sits behind it, OpenAI or Anthropic, that provider bills its own API calls and n8n adds nothing on top.

Two limits are worth knowing before the first import. Multitenancy, the way Weaviate isolates data per tenant inside one collection, has to be switched on at first ingestion with a Tenant Name, and it cannot be enabled or disabled afterwards. Search Filters and Metadata Keys are offered in the retrieval modes, not in Insert Documents, while Clear Data only shows up on Insert. The node sits at version 1, so an older workflow may show fewer options than the panel described here. The n8n review covers how the platform behaves once these AI workflows run daily.

Connect

How do you connect Weaviate to n8n?

  1. 01

    Create the Weaviate credential

    Open the Credentials menu in n8n and add a Weaviate credential, the stored set of secrets a node reuses without you retyping them. The node supports one authentication method, API key, and the credential is shared across every workflow that touches the same cluster. The Connection Type you pick next decides which fields the form asks for.

  2. 02

    Point it at Weaviate Cloud

    With Connection Type set to Weaviate Cloud, two values are needed: the Weaviate Cloud Endpoint of your database and the Weaviate Api Key. Both come from the cluster page in your Weaviate Cloud account. Weaviate offers a free sandbox, which is enough to build the collection and test a first insert before committing to anything bigger.

  3. 03

    Or describe your own cluster

    Custom Connection is for a Weaviate you deploy yourself. Alongside the Weaviate Api Key, fill Custom Connection HTTP Host and Custom Connection HTTP Port, 8080 by default, plus Custom Connection HTTP Secure for HTTPS. Weaviate also speaks gRPC, so Custom Connection gRPC Host, Custom Connection gRPC Port, 50051 by default, and Custom Connection gRPC Secure have to match your deployment.

Parameters

Which parameters does the node expose?

The Weaviate Vector Store node has 2 parameters. For each one: the node as you configure it in n8n, what the parameter changes, and our field notes.

01

Operation Mode

mode

What you see in n8n

Notes & use cases

This selector decides everything else on the panel: it tells the node whether the run stores documents or reads them.

Key parameters

  • Get Many (load): returns ranked documents for a query, driven by Prompt, Limit, Include Metadata and Rerank Results.
  • Insert Documents (insert): writes documents into the vector store.
  • Retrieve Documents (As Vector Store for Chain/Tool) (retrieve): exposes the store to other AI nodes.
  • Retrieve Documents (As Tool for AI Agent) (retrieve-as-tool): needs Name and Description.
Use cases
one workflow fills the collection on a schedule in Insert Documents mode, a second answers questions from it in Get Many mode.
02

Weaviate Collection

weaviateCollection

What you see in n8n

Notes & use cases

Every mode needs to know which collection it works on, and this required field is where that name goes. It is a resource locator, so the value can be chosen from the cluster or written as an expression when the collection depends on the item being processed.

Key parameters

  • Weaviate Collection: required in all four modes, typed directly or built from the incoming item with {{ $json.collection }}.
Use cases
an agency keeps one collection per client and resolves the name from the record that triggered the run, so the same workflow serves every account without duplication.
Need help

Need help automating Weaviate Vector Store with n8n?

A person reads every message.

FAQ

Weaviate Vector Store in n8n: common questions

01Is the n8n Weaviate Vector Store node included with n8n?
Yes. It ships inside n8n's AI package, @n8n/n8n-nodes-langchain, so nothing has to be installed and n8n charges nothing extra for it. That holds on n8n Cloud, the hosted offer run by n8n, and on a self-hosted instance under the Community Edition and its Sustainable Use license. A workflow behaves the same way in both places. Costs that do appear come from outside n8n: your Weaviate cluster, and the provider of whichever model generates the embeddings, which bills its own API calls under its own terms.
02What do you need to make it work?
A Weaviate credential, created once in the Credentials menu and reused everywhere. Authentication is done with an API key. If the cluster runs on Weaviate Cloud, the credential needs the Weaviate Cloud Endpoint and the Weaviate Api Key, and the free sandbox is enough to try it. For a cluster you host yourself, pick Custom Connection and fill Custom Connection HTTP Host and Port, 8080 by default, plus the gRPC Host and Port, 50051 by default, with the matching Secure switches when the endpoints are behind HTTPS. A GRPC Proxy option is available when gRPC traffic has to go through a proxy.
03What are the limits of the node?
The panel is deliberately small: two head parameters, Operation Mode and Weaviate Collection, with everything else living in Options. Some options only exist in some modes. Search Filters and Metadata Keys are available in Get Many and in both retrieve modes, never on insert, while Clear Data appears only in Insert Documents. Multitenancy is the strictest rule: a Tenant Name has to be passed at first ingestion, and multitenancy cannot be turned on or off after the collection is created. The node is at version 1, so an older workflow may expose fewer options than the current panel.
04What should you plug in first to get a working setup?
Start with Insert Documents and a real collection, because nothing can be retrieved before something is stored. A loader supplies the documents and their metadata, and Embedding Batch Size, 200 by default, controls how many get embedded per batch. Once the collection holds data, switch a second workflow to Get Many and check that Prompt and Limit return what you expect. Only then wire the agent version: Retrieve Documents (As Tool for AI Agent) with a Name and a Description precise enough that the model calls the tool at the right moment.
05n8n or Make for a Weaviate workflow?
It depends on where the data should live and how you want to pay. Make is hosted only, with no self-hosting option, and is billed per operation, which is predictable for light volumes. n8n can be self-hosted with Docker or npm, or used on n8n Cloud, and the same workflow runs identically either way. For a vector store, that choice usually follows the cluster: teams that host Weaviate themselves to keep documents in house tend to host n8n next to it. Both build workflows visually, so the deciding factors are hosting, data control and cost model rather than the editor.
Hack'celeration Lab

Get our weekly integration tips.

No spam. Unsubscribe anytime.

Same platform

Other docs you might need on this stack.

Browse all integrations