Skip to main content

​
Agent Examples

Ready-to-use agent recipes for the most common content and marketing workflows. Each example lists the exact steps to add, how to configure them, and what you’ll get out.

​
Example 1 — Blog Article Pipeline

What it does: Takes a topic, researches keywords and related questions, generates a content brief, and writes a full draft article. Steps:
  1. DataForSEO — keyword research
    • Keyword: {{topic}}
    • Limit: 25
  2. People Also Search For — find related angles
    • Query: {{topic}}
  3. Get Title Suggestions — decide the best angle
    • Topic: {{topic}}
    • Count: 10
    • Tone: informational
  4. Content Brief — structure the article
    • Topic: {{topic}}
    • Tone: informational
  5. Draft Article — write it
    • Brief: {{step_4.output}}
    • Word count: 2000
    • Tone: informational
Input fields to add: Topic Output: A 2,000-word article draft in Markdown, ready for editing and publishing.

​
Example 2 — AI Search Optimisation Article

What it does: Researches what AI assistants currently say about a topic using Perplexity, then writes a long-form article structured to win AI search answers. Steps:
  1. Perplexity Search — see what AI currently says
    • Prompt: What do people want to know about {{topic}}? What are the key questions and answers?
    • Model: Sonar Pro
  2. Google SERP Results — see who ranks today
    • Query: {{topic}}
    • Results: 10
  3. Content Brief — structure for AI search
    • Topic: Create a content brief for an article about {{topic}} optimised to rank in AI search engines like ChatGPT, Perplexity, and Google AI Overviews. Base the structure on this research: {{step_1.output}}
  4. Generate Article — write the full piece
    • Brief: {{step_3.output}}
    • Word count: 2500
    • AI Model: GPT-4o
Input fields to add: Topic Output: A long-form article built around the questions and answers that AI assistants surface for your topic.

​
Example 3 — Competitor Page Analysis

What it does: Crawls a competitor’s page, summarises their content angle, and suggests how you could write a better version. Steps:
  1. Get Web Content — crawl their page
    • Page URL: {{competitor_url}}
    • Extract main content: Yes
  2. Prompt LLM — analyse it
    • Prompt: Analyse this competitor article and identify: (1) their main angle and argument, (2) gaps and topics they missed, (3) weaknesses in their coverage. Article: {{step_1.output}}
    • Model: GPT-4o Mini
  3. Content Brief — write a better brief
    • Topic: Write a content brief for an article on the same topic as this competitor article that is significantly better. Competitor analysis: {{step_2.output}}
  4. Draft Article
    • Brief: {{step_3.output}}
    • Word count: 2000
Input fields to add: Competitor URL Output: An analysis of the competitor’s content and a draft article that improves on it.

​
Example 4 — Weekly SEO Report to Slack

What it does: Pulls your GSC keyword rankings and GA4 traffic data, summarises the highlights, and sends the summary to a Slack channel every week. Steps:
  1. Get GSC Rankings — pull keyword positions
    • Brand: (select your GSC connection)
    • Date range: 7 days
  2. Get GA4 Data — pull traffic metrics
    • Brand: (select your GA4 connection)
    • Date range: 7 days
  3. Prompt LLM — write the summary
    • Prompt: You are a marketing analyst. Write a concise weekly SEO performance summary (max 300 words) suitable for a Slack message. Cover: top performing pages, notable keyword movements, AI referral traffic trends. Data: GSC rankings: {{step_1.output}} GA4 traffic: {{step_2.output}}
    • Model: GPT-4o Mini
  4. Slack Notify — send it
    • Message: {{step_3.output}}
    • Brand: (select your Slack connection)
Input fields to add: None — schedule this agent to run weekly via the trigger settings. Output: A plain-English summary of your SEO performance delivered straight to your Slack channel.

​
Example 5 — Content to CMS Pipeline

What it does: Generates a finished article and publishes it as a draft directly to your Sanity CMS. Steps:
  1. Content Brief
    • Topic: {{topic}}
    • Target audience: {{audience}}
    • Tone: conversational
  2. Generate Article
    • Brief: {{step_1.output}}
    • Word count: 1500
    • AI Model: GPT-4o
  3. Wait for Approval — human review before publishing
    • Instructions: Review this article draft before it goes to Sanity. The topic is {{topic}}.
    • Attach: {{step_2.output}}
    • Timeout: 48 hours
  4. Sanity Write — publish to CMS
    • Content: {{step_2.output}}
    • Document type: post
    • Title: (AI will infer from the article)
Input fields to add: Topic, Audience Output: A published draft in Sanity, with a human approval gate before it lands in the CMS.

​
Example 6 — Knowledge Base Builder

What it does: Crawls a URL, extracts the content, and saves it to your knowledge base so future agents can reference it. Steps:
  1. Get Web Content
    • Page URL: {{url}}
    • Extract main content: Yes
  2. Prompt LLM — clean and summarise
    • Prompt: Summarise this page into a concise, well-structured reference document. Include key facts, claims, and data points. Source: {{step_1.output}}
    • Model: GPT-4o Mini
  3. Upload to Knowledge Base
    • Content: {{step_2.output}}
    • Knowledge base: (select your knowledge base)
Input fields to add: URL Output: The page’s content is stored in your knowledge base and available to all future agents.

​
Tips for building your own

  • Chain outputs. The most powerful agents pass one step’s output into the next. Use {{step_N.output}} in your prompts and configurations.
  • Start with research, end with content. A good pattern: research steps first (SERP, Perplexity, DataForSEO), then a brief step, then the article step.
  • Use Agentic Mode for exploration. When you’re not sure of the right steps, use Agentic Mode and let the AI figure out the path.
  • Add approval gates for anything customer-facing. Always put a Wait for Approval step before Slack notifications to customers or CMS publishes.