Generate SEO Seed Keywords Using AI

1. Workflow Overview

This workflow, titled "Generate SEO Seed Keywords Using AI", is designed to assist SEO strategists and marketers in generating a focused list of seed keywords tailored to an Ideal Customer Profile (ICP). By leveraging AI language models, it produces 15-20 seed keywords that are relevant to the ICP’s needs, pain points, goals, and search behavior. These keywords form a foundational element of an SEO strategy, particularly for B2B SaaS companies.

The workflow is logically divided into the following blocks:

1.1 Input Reception: Captures the ICP details via manual input.
1.2 Data Aggregation: Prepares and aggregates ICP data for AI processing.
1.3 AI Processing: Uses an AI Agent (Anthropic Chat Model) to generate seed keywords based on the ICP.
1.4 Output Handling: Parses AI output and routes it to an external database or storage.
1.5 Workflow Control & Information: Provides user guidance, notes, and reminders embedded as sticky notes for setup and usage.

2. Block-by-Block Analysis

1.1 Input Reception

Overview: This block receives the Ideal Customer Profile (ICP) details from the user manually and initiates the workflow execution.
Nodes Involved:
- When clicking ‘Test workflow’
- Set Ideal Customer Profile (ICP)
Node Details:
- When clicking ‘Test workflow’
  - Type: Manual Trigger
  - Role: Entry point for manual testing or running the workflow on demand.
  - Configuration: Default manual trigger with no parameters.
  - Inputs: None
  - Outputs: Connects to "Set Ideal Customer Profile (ICP)"
  - Failure Modes: None expected here.
- Set Ideal Customer Profile (ICP)
  - Type: Set Node
  - Role: Collects user-defined ICP parameters as strings detailing product, pain points, goals, current solutions, and expertise level.
  - Configuration: Contains placeholders to be replaced by users with their ICP data for the following fields:
    - product
    - pain points
    - goals
    - current solutions
    - expertise level
  - Key Expressions: Static string assignments with bold placeholders for user replacement.
  - Inputs: Trigger from "When clicking ‘Test workflow’"
  - Outputs: Passes structured ICP data downstream
  - Failure Modes: If left unchanged, the placeholder text may confuse the AI or generate irrelevant keywords.

1.2 Data Aggregation

Overview: Aggregates all ICP data items into a single unified data structure to prepare for AI processing.
Nodes Involved:
- Aggregate for AI node
Node Details:
- Aggregate for AI node
  - Type: Aggregate
  - Role: Combines all incoming ICP data items into one aggregated dataset.
  - Configuration: Set to aggregate all item data into a single composite output.
  - Inputs: Receives data from "Set Ideal Customer Profile (ICP)"
  - Outputs: Feeds aggregated data into the AI Agent node
  - Failure Modes: Data aggregation failure is unlikely but could occur if input schema changes or is empty.

1.3 AI Processing

Overview: Uses an AI agent powered by the Anthropic Chat Model to generate a list of SEO seed keywords based on the ICP data.
Nodes Involved:
- Anthropic Chat Model
- AI Agent
- Split Out
Node Details:
- Anthropic Chat Model
  - Type: Anthropic Chat Language Model Node (Langchain integration)
  - Role: Provides AI language generation capabilities as the backend for the AI agent.
  - Configuration: Default options, no custom parameters set.
  - Inputs: Connected as the language model for the "AI Agent" node.
  - Outputs: Provides raw AI completions to "AI Agent".
  - Version: 1.2
  - Failure Modes: API authentication failure, rate limiting, network errors, or model errors.
- AI Agent
  - Type: Langchain Agent Node
  - Role: Implements complex AI prompt logic, including instructions and input integration, to generate seed keywords from ICP data.
  - Configuration:
    - System message defines the AI as an expert SEO strategist.
    - Instructions include detailed rules for keyword generation (e.g., number of keywords, keyword format, search intent, specificity).
    - Input is dynamically injected ICP data aggregated from prior nodes.
    - Output is a JSON object with two keys: {thoughts:}, {answer:} containing an array of seed keywords.
  - Key Expressions: Uses n8n expression syntax to inject ICP data from aggregated JSON.
  - Inputs: Aggregated ICP data from "Aggregate for AI node"
  - Outputs: Passes AI-generated response to "Split Out" node
  - Version: 1.6
  - Failure Modes: Expression evaluation errors, API call failures, malformed AI output, or empty responses.
  - Notes: Critical node requiring valid AI API credentials (Anthropic/OpenAI).
- Split Out
  - Type: Split Out
  - Role: Extracts the output.answer field from the AI Agent’s JSON response array to isolate the generated seed keywords.
  - Configuration: Configured to split out the field output.answer.
  - Inputs: From "AI Agent"
  - Outputs: Routes keywords to the database connector node
  - Failure Modes: If the AI output JSON structure changes or is invalid, this node may fail.

1.4 Output Handling

Overview: Routes the final list of seed keywords to a user-configured data sink such as a database, Google Sheet, or Airtable.
Nodes Involved:
- Connect to your own database
Node Details:
- Connect to your own database
  - Type: No Operation (NoOp) placeholder node
  - Role: Indicates where users should connect their own database or data sink to store or further process the keyword list.
  - Configuration: None, serves as a placeholder for custom integration.
  - Inputs: Receives keyword array from "Split Out"
  - Outputs: None
  - Failure Modes: None internally; depends on user-implemented downstream integration.

1.5 Workflow Control & Information

Overview: Provides user instructions, notes on usage, costs, and setup reminders embedded as sticky notes.
Nodes Involved:
- Sticky Note
- Sticky Note11
- Sticky Note14
- Sticky Note15
- Sticky Note16
- Sticky Note13 (disabled)
- Sticky Note12 (disabled)
- Sticky Note17 (disabled)
Node Details:
- Sticky Note
  - Content: Overview of workflow purpose, outputs, and prerequisites.
  - Position: Top-left area to provide initial context.
- Sticky Note11
  - Content: Cost estimate for running the workflow using Claude Sonnet 3.5 model: Approx. $0.02-0.05 per run.
  - Position: Near AI processing nodes.
- Sticky Note14
  - Content: Reminder to connect to your own AI API credentials above.
  - Position: Near AI nodes.
- Sticky Note15
  - Content: Reminder to connect to your own database or storage solution to output generated keywords.
  - Position: Near the database connector.
- Sticky Note16
  - Content: Reminder to set the Ideal Customer Profile before running the workflow.
  - Position: Near the ICP input nodes.
- Sticky Note13, Sticky Note12, Sticky Note17
  - Status: Disabled and not active in current workflow execution. They contain brief notes on draft keyword generation, Airtable data formatting, and adding data to the database.

3. Summary Table

Node Name	Node Type	Functional Role	Input Node(s)	Output Node(s)	Sticky Note
When clicking ‘Test workflow’	Manual Trigger	Entry point to start workflow	None	Set Ideal Customer Profile (ICP)
Set Ideal Customer Profile (ICP)	Set	User input for ICP data	When clicking ‘Test workflow’	Aggregate for AI node	REQUIRED Set your Ideal Customer Profile before proceeding
Aggregate for AI node	Aggregate	Combines ICP data into one payload	Set Ideal Customer Profile (ICP)	AI Agent
Anthropic Chat Model	Anthropic Chat Model (Langchain)	AI language model backend	AI Agent (language model input)	AI Agent	REQUIRED Connect to your own AI API above
AI Agent	Langchain Agent	Generates seed keywords using AI	Aggregate for AI node	Split Out	REQUIRED Connect to your own AI API above
Split Out	Split Out	Extracts seed keywords from AI output	AI Agent	Connect to your own database
Connect to your own database	No Operation	Placeholder for database integration	Split Out	None	REQUIRED Connect to your own database / GSheet / Airtable base to output these
Sticky Note	Sticky Note	Workflow purpose and prerequisite notes	None	None	## Generate SEO Seed Keywords Using AI Pre-requisites: ICP, AI API account
Sticky Note11	Sticky Note	Cost estimate note	None	None	Costs to run Approx. $0.02-0.05 for a run using Claude Sonnet 3.5
Sticky Note14	Sticky Note	AI API connection reminder	None	None	REQUIRED Connect to your own AI API above
Sticky Note15	Sticky Note	Database connection reminder	None	None	REQUIRED Connect to your own database / GSheet / Airtable base to output these
Sticky Note16	Sticky Note	ICP setup reminder	None	None	REQUIRED Set your Ideal Customer Profile before proceeding
Sticky Note12 (disabled)	Sticky Note	Disabled - Airtable data formatting note	None	None
Sticky Note13 (disabled)	Sticky Note	Disabled - Draft seed keyword generation note	None	None
Sticky Note17 (disabled)	Sticky Note	Disabled - Add data to database note	None	None

4. Reproducing the Workflow from Scratch

Create Manual Trigger Node
- Type: Manual Trigger
- Purpose: Serve as the workflow start trigger for manual execution.
- No special configuration required.
Create Set Node: Set Ideal Customer Profile (ICP)
- Type: Set
- Configure fields (all type string) with placeholder text for:
  - product: "Replace this with a string detailing your intended product (if you have one)"
  - pain points: "Replace this with a string list of customer pain points"
  - goals: "Replace this with a string list of your customers key goals/objectives"
  - current solutions: "Replace this with a string detailing how your ideal customer currently solves their pain points"
  - expertise level: "Replace this with a string detailing customer level of expertise"
- Connect input from Manual Trigger.
Create Aggregate Node: Aggregate for AI node
- Type: Aggregate
- Set to aggregate all incoming items into one combined dataset.
- Connect input from Set Ideal Customer Profile node.
Create Langchain Anthropic Chat Model Node
- Type: Anthropic Chat Model (Langchain integration)
- Default options; no modifications necessary.
- No direct input connections; configured as language model for AI Agent node.
Create Langchain Agent Node: AI Agent
- Type: Langchain Agent
- Configure system message:
  "System: You are an expert SEO strategist tasked with generating 15-20 key head search terms (seed keywords) for a B2B SaaS company. Your goal is to create a comprehensive list of keywords that will attract and engage the ideal customer profile (ICP) described."
- Configure prompt text with detailed instructions and rules for generating keywords, including JSON format for output with {thoughts:}, {answer:} keys.
- Use n8n expression to inject ICP data from the aggregated input:
  {{ $json.data[0].product }} (and other ICP fields if necessary)
- Connect input from Aggregate for AI node.
- Set the Anthropic Chat Model node as the language model backend for this node.
- Ensure AI API credentials (Anthropic or OpenAI) are configured in n8n credentials.
Create Split Out Node: Split Out
- Type: Split Out
- Configure to extract the field: output.answer from the AI Agent output JSON.
- Connect input from AI Agent node.
Create No Operation Node: Connect to your own database
- Type: No Operation (NoOp)
- Placeholder node where users will connect their own database, Google Sheets, Airtable, or other data sinks.
- Connect input from Split Out node.
Create Sticky Notes (Optional but Recommended):
- Add sticky notes with the following contents positioned near relevant nodes:
  - Workflow purpose and prerequisites.
  - Reminder to set ICP before running.
  - Reminder to connect AI API credentials.
  - Reminder to connect the output to a database or storage.
  - Cost estimate for running AI models.

5. General Notes & Resources

Note Content	Context or Link
Made by Simon @ automake.io	https://automake.io
Workflow costs approx. $0.02-0.05 per run on Claude Sonnet 3.5 model	Cost estimate sticky note in workflow
Recommended AI API accounts: OpenAI or Anthropic	Setup prerequisites
Replace the NoOp node with your own database, Google Sheet, or Airtable integration to store generated keywords	User customization requirement

This documentation enables users and automation agents to fully understand, reproduce, and modify the "Generate SEO Seed Keywords Using AI" workflow while anticipating configuration needs and potential error points.

16 KiB Raw Blame History Unescape Escape