Resources · n8n integration

n8n Embeddings AWS Bedrock nodeConfigure Embeddings AWS Bedrock in n8n.

Embeddings turn text into numbers a machine can compare. The n8n Embeddings AWS Bedrock node hands that work to Amazon Bedrock, with 2 parameters to set and 2 authentication methods. It is a sub-node: it plugs into a node that needs embeddings and never runs on its own.

Try the Embeddings AWS Bedrock node on n8nSee the 2 parameters

Verified Trustpilot reviews · AI, automation & growth agency

Why automate

What does the n8n Embeddings AWS Bedrock node do?

The Embeddings AWS Bedrock node calls Amazon Bedrock to turn a piece of text into an embedding, a list of numbers that places the meaning of that text in a space where similar texts sit close together. It never works alone in a workflow. You attach it to a root node that asks for embeddings, and that node decides what gets embedded and when.

Start with the indexing side. A vector store node reads your support articles, product sheets or contract clauses, and asks this sub-node for the embedding of each chunk of text. A sub-node is a small node with no main input of its own: it hangs off a port of a bigger node and answers when that node calls it. Nothing is written to the store until Bedrock has answered, so a wrong model choice shows up as an empty index rather than a silent drift.

Then the search side. When a question arrives, the same node embeds the question, and the vector store compares that vector with the ones it already holds. Use the exact same model on both sides: a text embedded with one Bedrock model and searched with another gives results that look random. That retrieval step is what feeds grounded answers to an AI Agent, instead of letting it guess.

Third, the assistant case. A shop connects its catalog and its shipping policy to a vector store, the agent answers buyer questions from that content, and the embeddings node sits underneath as the piece that makes the catalog searchable by meaning rather than by keyword. Nobody sees it in the conversation. Everything depends on it.

When should you pick something else? This node only speaks to Amazon Bedrock, through an AWS credential. If your keys and your data already live somewhere else, an embeddings sub-node from that provider costs you less setup, and the OpenAI page covers that route. Bedrock earns its place when the account, the IAM policies and the region are already yours, and you want the vectors generated inside that perimeter.

The limits are worth knowing before you build. The node exposes 2 parameters, Authentication and Model, plus a short list of options: Additional Model Request Fields for model specific settings sent as JSON, Max Retries, and Timeout in milliseconds, which you set to 0 to disable. The node is at version 1, so an older workflow shows the same short panel. If you want the wider picture of where this fits, the n8n review and the n8n training pages go through the AI nodes as a whole.

Connect

What do you need to connect it to AWS?

  1. 01

    Pick the authentication method

    Open the node and look at Authentication. AWS (IAM) uses an IAM access key and asks for an AWS credential. AWS (Assume Role) temporarily assumes an IAM role and asks for an AWS (Assume Role) credential instead. The second one suits teams that already hand out roles rather than long lived keys.

  2. 02

    Create the credential once

    Credentials live in the n8n Credentials menu, not inside the workflow. Create the AWS credential there, and every workflow on that instance can reuse it, including a second embeddings node in another automation. A credential is simply the stored set of secrets n8n uses to sign its calls, so you never paste a key into a node parameter.

  3. 03

    Check the IAM permissions

    The Model dropdown is filled by AWS, not by n8n. It lists on-demand embedding models and embedding inference profiles together. If the list comes back empty or looks short, the IAM role behind the credential is probably missing bedrock:ListFoundationModels or bedrock:ListInferenceProfiles. Add the permission, then reopen the node.

Connections

What do you plug this node into?

n8n splits its AI nodes in two: a root node receives the workflow items through its main input, and sub-nodes attach to its ports, one type of sub-node per port. This one is a sub-node with a single output.

Output (what it plugs into)

  • Embeddingsai_embedding
01EmbeddingsThis is the only connection the node has. Its output plugs into the embeddings port of a root node that needs vectors, and the root node drives it: no main input, no standalone run, no schedule of its own.

This is the only connection the node has. Its output plugs into the embeddings port of a root node that needs vectors, and the root node drives it: no main input, no standalone run, no schedule of its own.

Key parameters

  • Required: a sub-node never executes alone, so the node stays idle until a root node calls it through this port.
Use cases
indexing a knowledge base once, then answering questions from it every day with the same model on both ends.
Parameters

Which parameters do you actually set?

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

01

Authentication

authentication

What you see in n8n

Notes & use cases

This selector decides how n8n proves to AWS that it is allowed to call Bedrock. Set it before anything else: the credential field right under it changes with your choice, and the model list will not load until the credential is valid.

Key parameters

  • AWS (IAM) (iam): uses an IAM access key through an AWS credential, the shortest path when you already have keys.
  • AWS (Assume Role) (assumeRole): temporarily assumes an IAM role through an AWS (Assume Role) credential, for accounts that grant access by role.
Use cases
a team that rotates roles instead of keys keeps the same workflow and only swaps this selector.
02

Model

model

What you see in n8n

Notes & use cases

Here you choose which Bedrock model generates the embedding. The dropdown mixes on-demand embedding models and embedding inference profiles, so the entry you pick is not always a bare model name. Whatever you choose becomes part of your index, and changing it later means embedding everything again.

Key parameters

  • Model (model): the model or inference profile that generates the embedding; switch the field to Expression mode and type the identifier directly when the dropdown stays empty or incomplete.
Use cases
pinning the identifier through an expression keeps a production workflow on a known model even if the dropdown behaves differently on another instance.
Need help

Need help automating Embeddings AWS Bedrock with n8n?

A person reads every message.

FAQ

Questions people ask next

01Is the Embeddings AWS Bedrock node free in n8n?
Yes on the n8n side. The node ships inside n8n's AI package, so there is nothing to install and nothing extra to pay, on n8n Cloud as well as on a self-hosted instance under the Community Edition and its Sustainable Use license. A workflow behaves the same either way. What does cost money is Amazon Bedrock itself: AWS bills the embedding calls your workflow makes, under its own terms, and n8n adds nothing on top of that. So the budget question belongs to your AWS account, not to your automation tool.
02What credentials does it need to work?
An AWS credential, chosen through the Authentication parameter. AWS (IAM) expects an AWS credential built on an IAM access key. AWS (Assume Role) expects an AWS (Assume Role) credential and temporarily assumes a role instead. You create either one once in the Credentials menu of n8n and reuse it across workflows. The IAM identity behind it also needs bedrock:ListFoundationModels and bedrock:ListInferenceProfiles for the Model dropdown to fill itself, and custom Bedrock Endpoint and Bedrock Runtime Endpoint values when Bedrock is reached through a VPC interface endpoint without private DNS.
03What are the limits of the n8n Embeddings AWS Bedrock node?
It does one thing: generate embeddings through Bedrock. It has no main input, no trigger, and no way to run by itself, so it only works attached to a root node that asks for vectors. The panel is short by design, with 2 parameters and the Additional Model Request Fields, Max Retries and Timeout options, and the node is at version 1. Model specific settings are not separate fields: you send them as JSON in Additional Model Request Fields, following what each model family accepts.
04What do you need to plug in to get it running?
At minimum, a root node with an embeddings port, plus a valid AWS credential on this node. A vector store is the natural starting point: it writes your documents as vectors during indexing and reads them back during a search. Add an agent above it only once retrieval returns sensible results, because an agent hides what the store actually found. Keep the same Model on the write side and the read side. Two different Bedrock models produce vectors that cannot be compared, and the search silently returns noise.
05n8n or Make for this kind of AI workflow?
It depends on where you want the data and how you want to pay. Make is hosted by Make, with no self-hosting option, and billed per operation, which is predictable for small volumes and less so when a workflow embeds thousands of chunks. n8n runs on n8n Cloud or on your own servers through Docker or npm, so the text you embed can stay inside your perimeter, next to the AWS account that answers. Both build workflows visually. Pick on hosting, data control and cost model rather than on the canvas.
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