January 22, 2026

OpenAI + Pinecone: chat requests into templates

Lisa Granqvist Partner Workflow Automation Expert

Get a free AI assessment → ⬇️ Use template

You get a one-line request in chat. You build something. Then the follow-up messages start. “Can you add search?” “We need memory.” “Actually, make it work with Gmail.” The back-and-forth is where timelines quietly die.

This hits automation agencies hardest, honestly, because every clarification is unbilled time. But a marketing ops lead trying to ship internal workflows feels it too, and so does a product-minded founder who just wants a working prototype. With OpenAI Pinecone templates, you turn messy chat requests into a ready-to-import n8n workflow file without hand-wiring nodes.

This workflow acts like a workflow “architect.” You’ll see what it automates, what you get out the other end, and what you need to run it reliably.

How This Automation Works

The full n8n workflow, from trigger to final output:

n8n Workflow Template: OpenAI + Pinecone: chat requests into templates

Click to explore

flowchart LR

    subgraph sg0["Manual Test Launcher Flow"]
        direction LR
        n2@{ icon: "mdi:vector-polygon", form: "rounded", label: "OpenAI Embedding Maker B", pos: "b", h: 48 }
        n3@{ icon: "mdi:robot", form: "rounded", label: "Standard Document Loader", pos: "b", h: 48 }
        n4@{ icon: "mdi:play-circle", form: "rounded", label: "Manual Test Launcher", pos: "b", h: 48 }
        n5["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/httprequest.dark.svg' width='40' height='40' /></div><br/>Fetch Crawl Output"]
        n6@{ icon: "mdi:swap-horizontal", form: "rounded", label: "Conditional Branch", pos: "b", h: 48 }
        n7["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/httprequest.dark.svg' width='40' height='40' /></div><br/>Retrieve Crawl Results"]
        n8@{ icon: "mdi:cog", form: "rounded", label: "Pause 30 Seconds", pos: "b", h: 48 }
        n9@{ icon: "mdi:cog", form: "rounded", label: "Pause 10 Seconds", pos: "b", h: 48 }
        n12@{ icon: "mdi:cube-outline", form: "rounded", label: "Pinecone Index Trainer", pos: "b", h: 48 }
        n20@{ icon: "mdi:cog", form: "rounded", label: "Convert Raw to File", pos: "b", h: 48 }
        n21@{ icon: "mdi:swap-vertical", form: "rounded", label: "Set Crawl Endpoint", pos: "b", h: 48 }
        n22@{ icon: "mdi:robot", form: "rounded", label: "Recursive Text Splitter", pos: "b", h: 48 }
        n6 --> n9
        n6 --> n20
        n8 --> n7
        n5 --> n8
        n21 --> n5
        n9 --> n7
        n7 --> n6
        n20 --> n12
        n2 -.-> n12
        n3 -.-> n12
        n22 -.-> n3
        n4 --> n21
    end

    subgraph sg1["Chat Message Intake Flow"]
        direction LR
        n0@{ icon: "mdi:play-circle", form: "rounded", label: "Chat Message Intake", pos: "b", h: 48 }
        n1@{ icon: "mdi:robot", form: "rounded", label: "Workflow Agent Core", pos: "b", h: 48 }
        n10@{ icon: "mdi:wrench", form: "rounded", label: "SerpAPI Search Tool", pos: "b", h: 48 }
        n11@{ icon: "mdi:cube-outline", form: "rounded", label: "Pinecone Retrieval Tool", pos: "b", h: 48 }
        n13@{ icon: "mdi:brain", form: "rounded", label: "OpenRouter Chat Engine", pos: "b", h: 48 }
        n14@{ icon: "mdi:vector-polygon", form: "rounded", label: "OpenAI Embedding Maker", pos: "b", h: 48 }
        n15@{ icon: "mdi:robot", form: "rounded", label: "OpenAI Validator", pos: "b", h: 48 }
        n16["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/code.svg' width='40' height='40' /></div><br/>JSON Extraction Script"]
        n17@{ icon: "mdi:swap-vertical", form: "rounded", label: "Map Extracted JSON", pos: "b", h: 48 }
        n18@{ icon: "mdi:swap-vertical", form: "rounded", label: "Assign Agent Preferences", pos: "b", h: 48 }
        n19@{ icon: "mdi:cog", form: "rounded", label: "Render JSON File", pos: "b", h: 48 }
        n16 --> n17
        n15 --> n16
        n10 -.-> n1
        n1 --> n15
        n17 --> n19
        n18 --> n1
        n14 -.-> n11
        n13 -.-> n1
        n11 -.-> n1
        n0 --> n1
    end

    %% Styling
    classDef trigger fill:#e8f5e9,stroke:#388e3c,stroke-width:2px
    classDef ai fill:#e3f2fd,stroke:#1976d2,stroke-width:2px
    classDef aiModel fill:#e8eaf6,stroke:#3f51b5,stroke-width:2px
    classDef decision fill:#fff8e1,stroke:#f9a825,stroke-width:2px
    classDef database fill:#fce4ec,stroke:#c2185b,stroke-width:2px
    classDef api fill:#fff3e0,stroke:#e65100,stroke-width:2px
    classDef code fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px
    classDef disabled stroke-dasharray: 5 5,opacity: 0.5
    class n4,n0 trigger
    class n3,n22,n1,n15 ai
    class n13 aiModel
    class n10 ai
    class n12,n11 ai
    class n2,n14 ai
    class n6 decision
    class n5,n7 api
    class n16 code
    classDef customIcon fill:none,stroke:none
    class n5,n7,n16 customIcon

The Problem: Chat Requests Create Hidden Rework

A chat message is a terrible spec. It’s short, ambiguous, and it usually mixes business goals with half-remembered tool names. So you do the responsible thing: you translate it into a workflow, guess the right nodes, wire it up, then run into the same wall every time. Node versions don’t match. Field names are slightly off. A connection is missing. The requester comes back with “one more change,” and suddenly you’re rebuilding the same skeleton again and again.

The friction compounds. Not because you can’t build workflows, but because the input is messy and the output has to be exact.

You spend about 1-2 hours per request just translating vague intent into the minimum set of nodes.
Small JSON mistakes fail import checks, which means you lose time debugging instead of delivering.
“Can you also add search?” turns into a redesign, not a tweak, because RAG and tool nodes change the whole shape.
Without a knowledge base of n8n docs and patterns, every new build starts from memory and Google.

The Solution: Turn One Message Into an Import-Ready Template

This workflow listens for an incoming chat request and routes it to an AI “builder” agent designed specifically for n8n templates. The agent doesn’t just guess. It pulls relevant n8n documentation from a Pinecone vector store, optionally crawls docs pages to refresh that knowledge, and uses web search when it needs examples or current references. Then it generates the smallest workable set of n8n nodes and connections, validates the output, extracts clean JSON, and finally returns a downloadable file you can import straight into n8n. You end up with a template that passes n8n’s import check, plus supporting assets that make the next request smarter.

The workflow starts when a chat message comes in through n8n’s chat trigger. From there, an AI Agent orchestrates retrieval (Pinecone), search (SerpAPI), and generation (OpenRouter GPT-4o) to draft the workflow template. Finally, OpenAI validates the structure, a small script extracts the JSON, and n8n renders it into a file for download.

What You Get: Automation vs. Results

What This Workflow Automates

Results You’ll Get

Capturing a one-line chat request and turning it into a structured build brief.
Pulling the right n8n docs and patterns from Pinecone instead of manual searching.
Using web search for examples, then merging them into a coherent node plan.
Validating and exporting a ready-to-import JSON template file automatically.

Most teams cut “first draft workflow” time down to about 10-15 minutes.
Fewer import failures because the template is checked before you download it.
Less rework when clients ask for RAG, memory, or search features.
A repeatable intake process, so requests don’t depend on one person’s expertise.
A growing internal knowledge base as your Pinecone index improves over time.

Example: What This Looks Like

Say you get 5 workflow requests a week, and each one takes about 2 hours to turn into a solid first draft (translate the request, pick nodes, wire, fix import issues). That’s roughly 10 hours weekly before anyone even tests the business logic. With this workflow, you drop the request into chat (about 2 minutes), let the agent run retrieval + search + generation (often around 10 minutes of waiting), then download the template.json file. You’re now spending your time reviewing and tailoring, not building from scratch, which usually saves you most of that 10 hours.

What You’ll Need

n8n instance (try n8n Cloud free)
Self-hosting option if you prefer (Hostinger works well)
OpenAI for embeddings and template validation
Pinecone to store and retrieve n8n documentation vectors
OpenRouter API Key (get it from your OpenRouter dashboard)

Skill level: Intermediate. You’ll connect credentials, tweak prompts, and verify the exported JSON imports cleanly.

Don’t want to set this up yourself? Talk to an automation expert (free 15-minute consultation).

How It Works

A chat request comes in. The Chat Message Intake node captures whatever the user typed, even if it’s just one line like “monitor support inbox and summarize tickets.”

Knowledge gets pulled in. The agent consults your Pinecone index for relevant n8n docs, and it can also use SerpAPI to find current examples or references when the request is niche.

The workflow template is generated and checked. The AI Agent drafts the node graph using OpenRouter (GPT-4o by default), then an OpenAI Validator double-checks structure, field names, and connectivity so the template won’t fail on import.

You receive a downloadable JSON file. A small extraction script slices clean JSON from the agent output, then n8n renders it to a file (template.json) you can download and import immediately.

You can easily modify the agent’s preferred tools to match your stack, so it defaults to the nodes you actually want. See the full implementation guide below for customization options.

Step-by-Step Implementation Guide

Step 1: Configure the Chat Trigger

Set up the inbound chat trigger that starts the agent workflow, and keep a manual trigger for testing crawl tasks.

Add Chat Message Intake as the primary trigger for incoming chat messages (no parameters required).
Keep Manual Test Launcher connected to Set Crawl Endpoint for manual test runs.
Optionally leave Flowpast Branding in the canvas as a documentation note (no runtime impact).

Step 2: Connect AI Services & Credentials

Attach the language model, search tool, and vector database services used by the agent, and ensure all required API credentials are configured.

Configure OpenRouter Chat Engine and connect it to Workflow Agent Core via the AI language model port. Credential Required: Connect your openRouterApi credentials.
Configure SerpAPI Search Tool and connect it to Workflow Agent Core via the AI tool port. Credential Required: Connect your serpApi credentials.
Configure Pinecone Retrieval Tool with Mode set to retrieve-as-tool, Tool Name set to n8n_documentation, and Tool Description set to This vectorstore contains some of n8n's technical documentations.. Credential Required: Connect your pineconeApi credentials.
Configure Pinecone Index Trainer with Mode set to insert. Credential Required: Connect your pineconeApi credentials.
Connect OpenAI Embedding Maker to Pinecone Retrieval Tool via the AI embedding port. Credential Required: Connect your openAiApi credentials.
Connect OpenAI Embedding Maker B to Pinecone Index Trainer via the AI embedding port. Credential Required: Connect your openAiApi credentials.
Configure OpenAI Validator with Model set to o4-mini-2025-04-16 and connect it after Workflow Agent Core. Credential Required: Connect your openAiApi credentials.

Tip: OpenAI Embedding Maker and OpenAI Embedding Maker B are connected as embeddings for the Pinecone tools—confirm their credentials are added on those embedding nodes, not on the Pinecone nodes.

Step 3: Set Up the Agent & Preferences

Define the agent’s default preferences and system instructions so it can build workflows based on user requests.

In Assign Agent Preferences, add assignments for:
vector database → Pinecone
chat model → Open Router
embedding → text-embedding-3-large
web search tool → SerpAPI
In Workflow Agent Core, confirm the System Message includes the preference expressions:
{{ $json['vector database'] }}
{{ $json['chat model'] }}
{{ $json.embedding }}
{{ $json['web search tool'] }}
Ensure Chat Message Intake and Assign Agent Preferences both connect to Workflow Agent Core before the validation step.

Step 4: Configure the Crawl & Retrieval Loop

Set the crawl URL and build the wait/retry loop that checks for completed crawl results.

In Set Crawl Endpoint, set URL to https://api.firecrawl.dev/v1/crawl/[YOUR_ID].
In Fetch Crawl Output, set URL to the expression {{ $json.URL }}.
Connect Fetch Crawl Output → Pause 30 Seconds → Retrieve Crawl Results to allow the crawl job to complete.
Route Retrieve Crawl Results into Conditional Branch, then wire the “not ready” path to Pause 10 Seconds and back to Retrieve Crawl Results to form the retry loop.
Route the “ready” path from Conditional Branch to Convert Raw to File for downstream indexing.

⚠️ Common Pitfall: Replace [YOUR_ID] in Set Crawl Endpoint with your real Firecrawl crawl ID, or the HTTP request will fail.

Step 5: Validate, Extract, and Package JSON Output

Validate the agent output, extract the JSON, and prepare files for storage and indexing.

In OpenAI Validator, ensure the second message content uses the expression {{ $json.output }} to validate the agent’s raw output.
In JSON Extraction Script, keep the provided JavaScript for extracting JSON from code fences.
In Map Extracted JSON, set Mode to raw and JSON Output to {{ $json.extractedJson }}.
In Render JSON File, set Mode to each, Operation to toJson, and Binary Property Name to =.
Connect Convert Raw to File → Pinecone Index Trainer for indexing of crawl results.
Connect Recursive Text Splitter → Standard Document Loader → Pinecone Index Trainer to enable document preprocessing before indexing.

Step 6: Test & Activate Your Workflow

Validate the full flow using manual triggers and then enable it for production usage.

Click Execute Workflow on Manual Test Launcher to test the crawl loop and Pinecone indexing path.
Send a test message to Chat Message Intake and confirm Workflow Agent Core produces output that passes through OpenAI Validator and Render JSON File.
Check for successful JSON extraction in Map Extracted JSON and a generated file in Render JSON File.
When everything looks correct, toggle the workflow to Active for production use.

🔒

Unlock Full Step-by-Step Guide

Get the complete implementation guide + downloadable template

Common Gotchas

Pinecone credentials and environment settings are easy to mix up. If retrieval suddenly returns nothing, check your Pinecone index name and API key in n8n credentials first.
If you’re using Wait nodes or external rendering, processing times vary. Bump up the wait duration if downstream nodes fail on empty responses.
Default prompts in AI nodes are generic. Add your brand voice early or you’ll be editing outputs forever.

Frequently Asked Questions

How long does it take to set up this OpenAI Pinecone templates automation?

About 30 minutes once your API keys are ready.

Do I need coding skills to automate OpenAI Pinecone templates?

No. You’ll mostly connect credentials and edit a couple of text fields for prompts and preferences.

Is n8n free to use for this OpenAI Pinecone templates workflow?

Yes. n8n has a free self-hosted option and a free trial on n8n Cloud. Cloud plans start at $20/month for higher volume. You’ll also need to factor in OpenAI, OpenRouter, Pinecone, and SerpAPI usage costs.

Where can I host n8n to run this automation?

Two options: n8n Cloud (managed, easiest setup) or self-hosting on a VPS. For self-hosting, Hostinger VPS is affordable and handles n8n well. Self-hosting gives you unlimited executions but requires basic server management.

Can I customize this OpenAI Pinecone templates workflow for my own internal docs instead of n8n docs?

Yes, and it’s one of the best upgrades. Swap the crawl/source in the Standard Document Loader and Set Crawl Endpoint nodes to point at your URLs or files, then keep the Recursive Text Splitter + OpenAI Embedding Maker feeding Pinecone Index Trainer. Common customizations include hard-coding preferred apps in Assign Agent Preferences, changing the LLM by replacing the OpenRouter Chat Engine, and tightening the validator instructions so it matches your team’s “import rules.”

Why is my OpenAI connection failing in this workflow?

Usually it’s an expired or wrong API key, or the OpenAI account doesn’t have access to the model you selected. Update the credential in n8n, then re-run a single test execution to confirm the Embedding Maker and Validator both succeed. If it fails only under load, you may be hitting rate limits, so slow down batching or reduce how many documents you embed in one run.

How many templates can this OpenAI Pinecone templates automation handle?

A lot, as long as your n8n execution limits and API quotas can keep up.

Is this OpenAI Pinecone templates automation better than using Zapier or Make?

For template generation, yes, most of the time. n8n is built for multi-step logic, looping, branching, and “agent + retrieval” patterns without turning into a fragile mess of mini-zaps. You also get the option to self-host, which matters when you’re running lots of internal requests or you don’t want per-task pricing. Zapier or Make can still be fine for lightweight triggers and notifications, but they’re not great for building and validating JSON artifacts. If you want a second opinion for your use case, Talk to an automation expert.

You set this up once, and your next “can you build a workflow for…” request stops being a mini project. The workflow handles the repetitive wiring and validation so you can focus on what actually matters: making it work for the business.

Need Help Setting This Up?

Our automation experts can build and customize this workflow for your specific needs. Free 15-minute consultation—no commitment required.

Lisa Granqvist

Workflow Automation Expert

Expert in workflow automation and no-code tools.

{
"id": "",
"meta": {
"instanceId": "120fc85b2e3fd9ab4e04ec246c52baa54ef33567b3c497e5d9fba7e9a0dfdf91",
"templateCredsSetupCompleted": null,
"templateId": null
},
"name": "Automated Agent Workflow Builder",
"tags": [],
"nodes": [
{
"id": "flowpast-topbar-5024",
"name": "Flowpast Branding",
"type": "n8n-nodes-base.stickyNote",
"position": [
0,
20
],
"parameters": {
"color": 7,
"width": 1342,
"height": 80,
"content": "## Flowpast.com | Automation Workflow Library\n**\ud83d\udcd6 Full tutorial & setup guide:** flowpast.com"
},
"typeVersion": 1
},
{
"id": "ee1c8a9a-e0fb-46d2-8274-140f5b297fba",
"name": "Chat Message Intake",
"type": "@n8n/n8n-nodes-langchain.chatTrigger",
"position": [
13,
144
],
"webhookId": "4e30031a-81f0-4d4d-bc44-2347530b51e8",
"parameters": {
"options": []
},
"typeVersion": 1.1
},
{
"id": "243ba102-72bf-441a-b247-d784b3b054bd",
"name": "Workflow Agent Core",
"type": "@n8n/n8n-nodes-langchain.agent",
"onError": "continueRegularOutput",
"position": [
188,
158
],
"parameters": {
"options": {
"systemMessage": "# General Instruction\nYou are a helpful AI agent that outputs n8n workflows in JSON format.\n\n# Input\nYou receive the user request describing what the agent should accomplish.\n\n# Output\nThe output must be a valid n8n template JSON.\n\n# Workflow\nThink carefully about the architecture. Populate the template with nodes and connect them properly. Use tools when helpful. Ensure all nodes except the trigger and final output are connected to at least one input and one output.\n\n# Tools\nYou have access to:\n1) Pinecone vector store containing some n8n documentation\n2) SerpAPI web search tool\n\n# User node preferences (use these if not specified by the user):\nvector database: {{ $json['vector database'] }}\nchat model: {{ $json['chat model'] }}\nembeddings: {{ $json.embedding }}\nweb search tool: {{ $json['web search tool'] }}\n\n# Examples\nExamples removed for brevity.\n\n# User request:\n"
}
},
"retryOnFail": true,
"typeVersion": 1.9,
"alwaysOutputData": false
},
{
"id": "e36992ff-fa9c-469c-a669-267bcdccae99",
"name": "OpenAI Embedding Maker B",
"type": "@n8n/n8n-nodes-langchain.embeddingsOpenAi",
"position": [
1230,
730
],
"parameters": {
"options": []
},
"credentials": {
"openAiApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 1.2
},
{
"id": "5e709878-fc9d-4ed3-bfbb-b019f9c287db",
"name": "Standard Document Loader",
"type": "@n8n/n8n-nodes-langchain.documentDefaultDataLoader",
"position": [
704,
732
],
"parameters": {
"options": []
},
"typeVersion": 1
},
{
"id": "f70e1fe1-618f-45c8-8379-f4742effb93b",
"name": "Manual Test Launcher",
"type": "n8n-nodes-base.manualTrigger",
"position": [
1186,
482
],
"parameters": [],
"typeVersion": 1
},
{
"id": "b34a7b23-f5b9-4046-9c92-a663698c0e92",
"name": "Fetch Crawl Output",
"type": "n8n-nodes-base.httpRequest",
"position": [
670,
486
],
"parameters": {
"url": "={{ $json.URL }}",
"options": []
},
"typeVersion": 4.2
},
{
"id": "ea2a97ef-f2bb-4934-a040-7fa96b918101",
"name": "Conditional Branch",
"type": "n8n-nodes-base.if",
"onError": "continueErrorOutput",
"position": [
728,
592
],
"parameters": {
"options": []
},
"typeVersion": 2.2
},
{
"id": "a34dea8f-fb4f-4c94-834d-eb5948cc7286",
"name": "Retrieve Crawl Results",
"type": "n8n-nodes-base.httpRequest",
"position": [
404,
638
],
"parameters": {
"options": []
},
"typeVersion": 4.2
},
{
"id": "63cd8437-92c9-4a33-aabf-3238b7b4a919",
"name": "Pause 30 Seconds",
"type": "n8n-nodes-base.wait",
"position": [
464,
430
],
"webhookId": "e4866198-b806-4746-b57f-97333c497dbf",
"parameters": [],
"typeVersion": 1.1
},
{
"id": "c6830132-8d74-4209-9d41-6c42c1fb7e66",
"name": "Pause 10 Seconds",
"type": "n8n-nodes-base.wait",
"position": [
932,
644
],
"webhookId": "6cc8c637-bafb-4d64-9cd4-8ae2a9f5b7ca",
"parameters": [],
"typeVersion": 1.1
},
{
"id": "95161389-9ea0-484b-abdf-82a8a0479b9a",
"name": "SerpAPI Search Tool",
"type": "@n8n/n8n-nodes-langchain.toolSerpApi",
"position": [
48,
306
],
"parameters": {
"options": []
},
"credentials": {
"serpApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 1
},
{
"id": "da8d5a8a-c641-453f-a7bb-817099e8592f",
"name": "Pinecone Retrieval Tool",
"type": "@n8n/n8n-nodes-langchain.vectorStorePinecone",
"position": [
216,
442
],
"parameters": {
"mode": "retrieve-as-tool",
"options": [],
"toolName": "n8n_documentation",
"pineconeIndex": {
"__rl": true,
"mode": "list",
"value": "agentbuilder",
"cachedResultName": "agentbuilder"
},
"toolDescription": "This vectorstore contains some of n8n's technical documentations."
},
"credentials": {
"pineconeApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 1.1
},
{
"id": "0838fd0e-8e1b-4c37-8c31-8ab9306fff89",
"name": "Pinecone Index Trainer",
"type": "@n8n/n8n-nodes-langchain.vectorStorePinecone",
"position": [
936,
786
],
"parameters": {
"mode": "insert",
"options": [],
"pineconeIndex": {
"__rl": true,
"mode": "list",
"value": "agentbuilder",
"cachedResultName": "agentbuilder"
}
},
"credentials": {
"pineconeApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 1.1
},
{
"id": "b4bca1b1-80bc-42fa-b434-00eabd926aff",
"name": "OpenRouter Chat Engine",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenRouter",
"position": [
200,
292
],
"parameters": {
"options": []
},
"credentials": {
"openRouterApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 1
},
{
"id": "e9f693e5-4733-4987-a7a8-7f6b7823b221",
"name": "OpenAI Embedding Maker",
"type": "@n8n/n8n-nodes-langchain.embeddingsOpenAi",
"position": [
534,
388
],
"parameters": {
"options": []
},
"credentials": {
"openAiApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 1.2
},
{
"id": "ebe6227a-4f12-44d0-ae8c-516e89e31f9f",
"name": "OpenAI Validator",
"type": "@n8n/n8n-nodes-langchain.openAi",
"position": [
540,
120
],
"parameters": {
"modelId": {
"__rl": true,
"mode": "list",
"value": "o4-mini-2025-04-16",
"cachedResultName": "O4-MINI-2025-04-16"
},
"options": [],
"messages": {
"values": [
{
"role": "system",
"content": "Transform the input into valid n8n template JSON format. make sure all nodes are correctly set up and CONNECTED!\n\nmake especially sure the AI agent(s) are correctly set up and connected. here is an example:\n\n{\n  \"id\": \"ai-agent-gpt4o-serpapi-manual\",\n  \"name\": \"GPT-4o Agent with SerpAPI & Memory (Manual Trigger)\",\n  \"nodes\": [\n  {\n  \"id\": \"dba53fd3-6bf0-46c9-9659-286f643fd6d1\",\n  \"name\": \"Manual Trigger\",\n  \"type\": \"n8n-nodes-base.manualTrigger\",\n  \"position\": [\n  -420,\n  -40\n  ],\n  \"parameters\": {},\n  \"typeVersion\": 1\n  },\n  {\n  \"id\": \"dd16c4ff-c7d4-45fc-b75a-29bdb6dcc0c2\",\n  \"name\": \"Window Buffer Memory\",\n  \"type\": \"@n8n/n8n-nodes-langchain.memoryBufferWindow\",\n  \"position\": [\n  -140,\n  -40\n  ],\n  \"parameters\": {},\n  \"typeVersion\": 1.3\n  },\n  {\n  \"id\": \"32a1cc05-dab8-4be1-bd6a-8396fbe0df0e\",\n  \"name\": \"GPT-4o Chat Model\",\n  \"type\": \"@n8n/n8n-nodes-langchain.lmChatOpenAi\",\n  \"position\": [\n  -140,\n  -220\n  ],\n  \"parameters\": {\n  \"model\": {\n  \"__rl\": true,\n  \"mode\": \"list\",\n  \"value\": \"gpt-4o\",\n  \"cachedResultName\": \"gpt-4o\"\n  },\n  \"options\": {}\n  },\n  \"credentials\": {\n  \"openAiApi\": {\n  \"id\": \"<OPENROUTER_CREDENTIAL_ID>\",\n  \"name\": \"OpenRouter API Key\"\n  }\n  },\n  \"typeVersion\": 1.2\n  },\n  {\n  \"id\": \"ea7379d9-b6dd-49ff-ac01-2d9c608f7c11\",\n  \"name\": \"SerpAPI\",\n  \"type\": \"@n8n/n8n-nodes-langchain.toolSerpApi\",\n  \"position\": [\n  -140,\n  140\n  ],\n  \"parameters\": {\n  \"options\": {}\n  },\n  \"credentials\": {\n  \"serpApi\": {\n  \"id\": \"<SERPAPI_CREDENTIAL_ID>\",\n  \"name\": \"SerpAPI Key\"\n  }\n  },\n  \"typeVersion\": 1\n  },\n  {\n  \"id\": \"10a3f41d-4f1a-4efa-92b3-7adb3f98f604\",\n  \"name\": \"AI Agent\",\n  \"type\": \"@n8n/n8n-nodes-langchain.agent\",\n  \"position\": [\n  160,\n  -40\n  ],\n  \"parameters\": {\n  \"text\": \"=Respond to the user\u2019s request given in the trigger item.\",\n  \"options\": {\n  \"systemMessage\": \"You are an AI assistant that can perform web searches with SerpAPI when helpful.\"\n  },\n  \"promptType\": \"define\"\n  },\n  \"typeVersion\": 1.7\n  },\n  {\n  \"id\": \"f8b0c3ae-8423-4db9-a287-3e3c640ef0bb\",\n  \"name\": \"Set Response\",\n  \"type\": \"n8n-nodes-base.set\",\n  \"position\": [\n  400,\n  -40\n  ],\n  \"parameters\": {\n  \"assignments\": {\n  \"assignments\": [\n  {\n  \"id\": \"acb7e4d0-f718-4eae-8f50-0ca15c5c68e2\",\n  \"name\": \"answer\",\n  \"type\": \"string\",\n  \"value\": \"={{ $json.output }}\"\n  }\n  ]\n  },\n  \"options\": {}\n  },\n  \"typeVersion\": 3.4\n  }\n  ],\n  \"connections\": {\n  \"Manual Trigger\": {\n  \"main\": [\n  [\n  {\n  \"node\": \"Window Buffer Memory\",\n  \"type\": \"main\",\n  \"index\": 0\n  }\n  ]\n  ]\n  },\n  \"Window Buffer Memory\": {\n  \"main\": [\n  [\n  {\n  \"node\": \"AI Agent\",\n  \"type\": \"main\",\n  \"index\": 0\n  }\n  ]\n  ],\n  \"ai_memory\": [\n  [\n  {\n  \"node\": \"AI Agent\",\n  \"type\": \"ai_memory\",\n  \"index\": 0\n  }\n  ]\n  ]\n  },\n  \"GPT-4o Chat Model\": {\n  \"ai_languageModel\": [\n  [\n  {\n  \"node\": \"AI Agent\",\n  \"type\": \"ai_languageModel\",\n  \"index\": 0\n  }\n  ]\n  ]\n  },\n  \"SerpAPI\": {\n  \"ai_tool\": [\n  [\n  {\n  \"node\": \"AI Agent\",\n  \"type\": \"ai_tool\",\n  \"index\": 0\n  }\n  ]\n  ]\n  },\n  \"AI Agent\": {\n  \"main\": [\n  [\n  {\n  \"node\": \"Set Response\",\n  \"type\": \"main\",\n  \"index\": 0\n  }\n  ]\n  ]\n  }\n  },\n  \"active\": true,\n  \"settings\": {\n  \"executionOrder\": \"v1\"\n  }\n  }\n"
},
{
"content": "={{ $json.output }}"
}
]
}
},
"credentials": {
"openAiApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 1.8
},
{
"id": "afcb27aa-257f-4c9a-8cb2-d54f23123443",
"name": "JSON Extraction Script",
"type": "n8n-nodes-base.code",
"position": [
714,
160
],
"parameters": {
"jsCode": "const items = $input.all();\nconst extractedJsons = items.map((item) => {\n  const content = item?.json?.message?.content;\n  const jsonStart = content.indexOf(\"```json\") + 7;\n  const jsonEnd = content.indexOf(\"```\", jsonStart);\n  const jsonContent = content.slice(jsonStart, jsonEnd);\n  try {\n    const parsedJson = JSON.parse(jsonContent);\n    return { extractedJson: parsedJson };\n  } catch (error) {\n    console.log(\"Failed to parse JSON:\", error);\n    return { extractedJson: null };\n  }\n});\nreturn extractedJsons;\n"
},
"typeVersion": 2
},
{
"id": "a802de97-238c-4f7f-9024-43d3fb14f7a8",
"name": "Map Extracted JSON",
"type": "n8n-nodes-base.set",
"position": [
1006,
176
],
"parameters": {
"mode": "raw",
"options": [],
"jsonOutput": "=\n {{ $json.extractedJson }}"
},
"typeVersion": 3.4
},
{
"id": "d67d0970-0f58-41f4-8d48-077247eb4b0a",
"name": "Assign Agent Preferences",
"type": "n8n-nodes-base.set",
"position": [
0,
336
],
"parameters": {
"options": [],
"assignments": {
"assignments": [
{
"id": "6905b43a-c181-4c14-a746-c4d82325cccf",
"name": "vector database",
"type": "string",
"value": "Pinecone"
},
{
"id": "cfe18b2f-051d-492d-bb63-dd91076bce68",
"name": "chat model",
"type": "string",
"value": "Open Router"
},
{
"id": "bb2d75ad-9544-4d48-b9d1-ddc2e5a124c1",
"name": "embedding",
"type": "string",
"value": "text-embedding-3-large"
},
{
"id": "1a763e6b-22df-477f-941b-4933231d636a",
"name": "web search tool",
"type": "string",
"value": "SerpAPI"
}
]
}
},
"typeVersion": 3.4
},
{
"id": "392f160f-5ffd-4799-a63d-adc240b7f26a",
"name": "Render JSON File",
"type": "n8n-nodes-base.convertToFile",
"position": [
1178,
148
],
"parameters": {
"mode": "each",
"options": {
"format": true,
"encoding": "utf8",
"fileName": ""
},
"operation": "toJson",
"binaryPropertyName": "="
},
"typeVersion": 1.1
},
{
"id": "efc08c98-cb77-45f0-b824-de2a22dee2fc",
"name": "Convert Raw to File",
"type": "n8n-nodes-base.convertToFile",
"position": [
1242,
586
],
"parameters": {
"options": []
},
"typeVersion": 1.1
},
{
"id": "9ff243d3-1871-4634-b5a6-dd9a745a7a84",
"name": "Set Crawl Endpoint",
"type": "n8n-nodes-base.set",
"position": [
992,
432
],
"parameters": {
"options": [],
"assignments": {
"assignments": [
{
"id": "60871e95-afba-4ff9-ba56-315ae39cd6f9",
"name": "URL",
"type": "string",
"value": "https://api.firecrawl.dev/v1/crawl/[YOUR_ID]"
}
]
}
},
"typeVersion": 3.4
},
{
"id": "03e4cd15-b458-4601-a9d4-c365a27e846e",
"name": "Recursive Text Splitter",
"type": "@n8n/n8n-nodes-langchain.textSplitterRecursiveCharacterTextSplitter",
"position": [
408,
790
],
"parameters": {
"options": []
},
"typeVersion": 1
}
],
"active": false,
"pinData": [],
"settings": {
"executionOrder": "v1"
},
"versionId": "",
"connections": {
"Conditional Branch": {
"main": [
[
{
"node": "Pause 10 Seconds",
"type": "main",
"index": 0
}
],
[
{
"node": "Convert Raw to File",
"type": "main",
"index": 0
}
]
]
},
"JSON Extraction Script": {
"main": [
[
{
"node": "Map Extracted JSON",
"type": "main",
"index": 0
}
]
]
},
"OpenAI Validator": {
"main": [
[
{
"node": "JSON Extraction Script",
"type": "main",
"index": 0
}
]
]
},
"Pause 30 Seconds": {
"main": [
[
{
"node": "Retrieve Crawl Results",
"type": "main",
"index": 0
}
]
]
},
"Fetch Crawl Output": {
"main": [
[
{
"node": "Pause 30 Seconds",
"type": "main",
"index": 0
}
]
]
},
"SerpAPI Search Tool": {
"ai_tool": [
[
{
"node": "Workflow Agent Core",
"type": "ai_tool",
"index": 0
}
]
]
},
"Set Crawl Endpoint": {
"main": [
[
{
"node": "Fetch Crawl Output",
"type": "main",
"index": 0
}
]
]
},
"Workflow Agent Core": {
"main": [
[
{
"node": "OpenAI Validator",
"type": "main",
"index": 0
}
]
]
},
"Pause 10 Seconds": {
"main": [
[
{
"node": "Retrieve Crawl Results",
"type": "main",
"index": 0
}
]
]
},
"Retrieve Crawl Results": {
"main": [
[
{
"node": "Conditional Branch",
"type": "main",
"index": 0
}
]
]
},
"Map Extracted JSON": {
"main": [
[
{
"node": "Render JSON File",
"type": "main",
"index": 0
}
]
]
},
"Convert Raw to File": {
"main": [
[
{
"node": "Pinecone Index Trainer",
"type": "main",
"index": 0
}
]
]
},
"Assign Agent Preferences": {
"main": [
[
{
"node": "Workflow Agent Core",
"type": "main",
"index": 0
}
]
]
},
"OpenAI Embedding Maker": {
"ai_embedding": [
[
{
"node": "Pinecone Retrieval Tool",
"type": "ai_embedding",
"index": 0
}
]
]
},
"OpenAI Embedding Maker B": {
"ai_embedding": [
[
{
"node": "Pinecone Index Trainer",
"type": "ai_embedding",
"index": 0
}
]
]
},
"Render JSON File": {
"main": [
[]
]
},
"Standard Document Loader": {
"ai_document": [
[
{
"node": "Pinecone Index Trainer",
"type": "ai_document",
"index": 0
}
]
]
},
"OpenRouter Chat Engine": {
"ai_languageModel": [
[
{
"node": "Workflow Agent Core",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"Pinecone Retrieval Tool": {
"ai_tool": [
[
{
"node": "Workflow Agent Core",
"type": "ai_tool",
"index": 0
}
]
]
},
"Chat Message Intake": {
"main": [
[
{
"node": "Workflow Agent Core",
"type": "main",
"index": 0
}
]
]
},
"Recursive Text Splitter": {
"ai_textSplitter": [
[
{
"node": "Standard Document Loader",
"type": "ai_textSplitter",
"index": 0
}
]
]
},
"Manual Test Launcher": {
"main": [
[
{
"node": "Set Crawl Endpoint",
"type": "main",
"index": 0
}
]
]
}
}
}