Resources · n8n integration

Perplexity n8n integrationAutomate Perplexity with n8n.

Need sourced web answers inside a workflow? The Perplexity n8n integration fits in a single node: 5 operations spread across 4 resources, from Sonar chat completions to embeddings and raw web search. Nothing on the Perplexity side starts a run, so the workflow begins elsewhere.

Try the Perplexity integration on n8nSee the 5 operations

Verified Trustpilot reviews · AI, automation & growth agency

Why automate

What does the Perplexity n8n integration actually do?

The Perplexity node ships with n8n and exposes 5 operations across 4 resources: Chat, Agent, Search and Embedding. It calls the Perplexity API with your own API key, so a workflow can put a question to a Sonar model, hand a multi-step task to an agent, pull ranked web results, or turn text into vectors. Nothing gets installed, and the same credential works on n8n Cloud and on a self-hosted instance.

Three jobs come up again and again. The first is a morning market brief: a Schedule Trigger fires, chat.complete asks a Sonar model what moved in your category, and the answer lands in Slack with its citations attached. If the brief belongs in an inbox instead, Gmail takes the delivery step.

The second is lead research. A new row shows up, search.search pulls ranked results about the company, and the summary goes back to Google Sheets next to the contact. The third is retrieval: embedding.createEmbedding turns product docs into vectors for a knowledge base, and embedding.createContextualized does the same for long documents where a chunk on its own means very little.

Reach for the HTTP Request node when the API exposes something this node does not. It calls any Perplexity endpoint and reuses the same credential through predefined authentication, so there is no second key to manage.

Two limits are worth knowing before you build. Perplexity has no trigger node in n8n, which means nothing on its side can start a workflow: that job goes to a Schedule Trigger, an n8n Webhook (a URL another service calls when something happens), or the trigger of a tool already in the picture. And a few options depend on the usage tier of the Perplexity API account: Return Images and Return Related Questions need Tier-2, while Search Domain Filter needs Tier-3 and accepts 3 domains at most.

If the open question is the platform rather than the node, the n8n review goes through the criteria. The short version here: the node itself costs nothing on the n8n side, and what gets consumed is consumed against your own Perplexity API key.

Connect

How do you connect Perplexity to n8n?

  1. 01

    Generate a Perplexity API key

    Sign in to your Perplexity account and create an API key, following the steps in the Perplexity API getting started guide. That key is the only thing this integration asks for: no OAuth screen, no scope to tick, no app to register on the Perplexity side. Copy it before you leave the page and keep it somewhere safe.

  2. 02

    Store the credential in n8n

    In n8n, open the Credentials menu and add a new Perplexity credential, the saved connection a node reuses to authenticate itself. It has one field to fill, the Perplexity API key. Once saved, it shows up in the dropdown of every Perplexity node in the instance, and rotating the key later is a change you make in one place.

  3. 03

    Pick a resource and run the node once

    Drop a Perplexity node into a workflow, select the credential, then choose the resource (Chat, Agent, Search or Embedding) and its operation. Run the node on its own with a short prompt before wiring anything upstream. The output panel shows the exact shape of the response, which is what you reference further down with an expression such as {{ $json.field }}.

Actions

What can the Perplexity node do?

The Perplexity node exposes 5 operations across 4 resources. For each one: the node as you configure it in n8n, the required fields, and our field notes.

Resources × operations matrix
ResourceMessage a ModelCreate Contextualized EmbeddingCreate EmbeddingCreate ResponseSearch
Agent✓
Chat✓
Embedding✓✓
Search✓

Agent

1 operation
01

Create a response

agent.createResponse

What you see in n8n

Notes & use cases

The Agent resource hands a whole task to Perplexity rather than a single question: it chains its own steps, searches and uses tools before writing the final response.

Key parameters

  • Input: the text prompt the agent starts from, often {{ $json.field }} carried in by the node before.
  • Model: the model to run, in provider/model form. Leave it empty when a preset already holds the setup.
  • Preset: the name of a saved configuration. Use a preset or a model, never both.
  • Max Steps: caps the agentic steps between 1 and 10, so a long task stays bounded.
Use cases
a piece of research that needs several passes before it is usable, kept short by Max Steps.

Chat

1 operation
02

Message a model

chat.complete

What you see in n8n

Notes & use cases

Sends a conversation to a Sonar model and brings back the completion together with the citations behind it.

Key parameters

  • Model: one of sonar, sonar-pro, sonar-reasoning-pro and sonar-deep-research.
  • Messages: each entry carries a Role and its Text; system messages come first, then user and assistant alternate.
  • Search Recency Filter: restricts the sources to the last Hour, Day, Week, Month or Year.
Use cases
a recurring brief where the model answers one fixed question and the recency filter keeps stale coverage out.

Embedding

2 operations
03

Create an embedding

embedding.createEmbedding

What you see in n8n

Notes & use cases

Converts text into vectors, the numeric form a vector database needs in order to retrieve a passage by meaning instead of by exact word.

Key parameters

  • Input Texts: the text to embed, one text per line; every line comes back as its own vector.
  • Model: pplx-embed-v1-0.6b or pplx-embed-v1-4b.
  • Dimensions: shortens the output vector; left at 0 or unset, the full model size is used.
Use cases
indexing a help center article by article, so a support agent finds the right page from a customer's own wording.
04

Create a contextualized embedding

embedding.createContextualized

What you see in n8n

Notes & use cases

Embeds the chunks of a document while keeping the rest of that document in view, so a paragraph that only says "the same applies here" still carries what it refers to.

Key parameters

  • Input Documents: a JSON array of documents, each document being an array of paragraph strings.
  • Model: pplx-embed-context-v1-4b, the single choice on this operation.
  • Encoding Format: Base64 Int8 or Base64 Binary; float is not supported on this endpoint.
Use cases
a long contract split into paragraphs, where each chunk has to stay attached to the clause around it.
Need help

Need help automating Perplexity with n8n?

A person reads every message.

FAQ

Perplexity and n8n, the questions that come next

01Is the Perplexity n8n integration free to use?
On the n8n side, yes. The Perplexity node ships with n8n at no extra cost, on n8n Cloud and on a self-hosted instance under the Community Edition and its Sustainable Use licence, and there is nothing to install before the node appears in the editor. What happens on the Perplexity side is a separate matter: every call goes through your own API key and follows whatever plan that key belongs to, and this page does not cover Perplexity pricing. One detail that trips people up: a handful of node options depend on the usage tier of that account, not on the n8n plan you run.
02What credentials do you need to connect Perplexity to n8n?
One API key, and that is the whole list. The Perplexity credential in n8n supports a single authentication method, an API key generated from a Perplexity account, so there is no OAuth screen, no scope list and no app to register first. Create the credential once in the Credentials menu and every workflow in the instance can pick it from the node dropdown. The same credential also serves the HTTP Request node through predefined authentication, which matters as soon as you need an endpoint the dedicated node does not expose. Rotating the key later means editing one credential instead of every workflow that calls Perplexity.
03What are the limits of the Perplexity node in n8n?
Three are worth planning around. Coverage first: the node exposes 5 operations across 4 resources, so anything outside chat completions, agent responses, web search and embeddings goes through the HTTP Request node with the same credential. Then the account tier: Return Images and Return Related Questions require Perplexity API usage Tier-2, and Search Domain Filter requires Tier-3 and takes 3 domains at most. Last, search volume: the Search operation returns up to 20 results in one call and its own Search Domain Filter accepts up to 20 domains, so a wide sweep means several calls rather than one.
04Does Perplexity trigger n8n workflows in real time?
No, and there is no in-between answer here: Perplexity has no trigger node in n8n, so nothing on its side can start a run. The workflow starts elsewhere and calls Perplexity as one of its steps. A Schedule Trigger covers a recurring brief. An n8n Webhook, a URL another service calls when something happens on its side, covers event-driven runs. The trigger of a tool already in the workflow covers the rest, a new row, a new message, a new email. In practice that fits how the tool works, since Perplexity answers questions it is asked rather than emitting events of its own.
05n8n or Make for Perplexity?
It depends on where the workflow should live and how you want to pay for it. Make is hosted only, with no self-hosting option, and it bills per operation, so a workflow that queries Perplexity all day grows in cost with volume. n8n runs on n8n Cloud, the hosted offer, or on your own servers through Docker or npm, and the workflow is the same either way, which keeps prompts, answers and the API key on infrastructure you control when that matters. Make leans more visual for linear scenarios; n8n gets closer to code once branching and expressions pile up. Pick on hosting, data control, cost model and how visual you want the logic.
Same platform

Other docs you might need on this stack.

Browse all integrations