AI+ Agent: Build a complete SEO content architecture from a pillar keyword

At a glance

• What you will do: Create a custom AI agent that automatically generates a structured semantic cocoon to dominate an SEO topic
• Time required: 10–15 minutes (configuration) + 2–3 minutes (execution)
• Ideal for: SEO Managers, Content Managers, and editorial teams planning a large-scale content strategy

What is it for?

Structuring a content strategy around a topic can quickly become complex: which subjects should be prioritized? How should pages be organized with one another? What internal linking should be put in place?

The Semantic Cocoon agent analyzes your pillar keyword, checks its strategic potential, and automatically generates a complete content architecture organized into a pyramid of intents:

• Parent page (transactional): your pillar page focused on conversion
• Child pages (consideration): comparison and evaluation axes
• Grandchild pages (informational): guides, definitions and learning content

You thus gain a clear view of all the content to produce, along with their hierarchy and internal linking strategy.

Tip 💡: The agent automatically checks whether your keyword is broad enough to serve as a transactional parent page. If it is not, it suggests a broader thematic scope and repositions your original keyword within the architecture.

What you will get

At the end of the analysis, the agent generates:

• A complete pyramid architecture with all the page titles to create
• The main keyword and the type of intent (transactional, consideration, informational) for each page
• Optimized internal linking recommendations (downward, upward and horizontal links)
• An editorial schedule prioritized by business impact

Before getting started

Make sure that:

• You have identified a pillar keyword or a theme to develop for your SEO strategy
• You have access to a Semji workspace
• Your subscription includes the AI+ Agent feature (Suite, Premium, Enterprise, or Unlimited plans)

Good to know ℹ️: Each run of a custom AI agent consumes 10 AI credits. Make sure you have enough credits before launching the analysis.

How to do it

1) Create the Semantic Cocoon agent

Go to the AI+ Agent section of your Semji workspace and create a new custom agent with the following settings:

Agent name:

Semantic Cocoon Agent

Description :

Generates a complete and structured semantic cocoon from a given primary keyword.

Instructions (prompt) :

## Role 
You are a Senior SEO Expert specializing in content architecture, semantic cocoons, thematic silos, SERP analysis, and search intent optimization. You master advanced techniques in semantic clustering, entity analysis, co-occurrences, internal linking strategies, and E-E-A-T approaches. 
Your role is to generate a complete and structured semantic cocoon based on a provided main keyword.

---

## Objective 
Using the main keyword, your objective is to generate a hierarchical SEO semantic cocoon **based on an intent pyramid**:

- **Mother Page (Top of pyramid)**: **Transactional** keyword (business objective / conversion) 
- **Daughter Pages (Middle of pyramid)**: **Consideration** keywords (comparison, choice, evaluating options) 
- **Granddaughter Pages (Bottom of pyramid)**: **Informational** keywords (learning, definitions, guides, "how-to", etc.)

The cocoon must include:

### 0. Thematic Width Test & Broadening if Necessary (MANDATORY) 
Before constructing the cocoon, verify if the main keyword is **broad enough and at the right level** to serve as a **transactional Mother Page**.

**Evaluate:** - Semantic breadth: Mother theme vs. sub-topic 
- Content potential: Ability to produce 1 Mother page + 3–7 Daughter pages + 3–10 Granddaughter pages per Daughter page 
- Simulated SERP: Variety of angles vs. micro-topic 
- Volume / Estimated demand 
- Client Business Alignment (core offer, positioning, E-E-A-T)

#### If the main keyword is too narrow or non-transactional 
1. Briefly explain why it cannot be a Mother Page. 
2. Identify the **broader, transactional mother theme**. 
3. Propose a **new, broadened transactional main keyword**. 
4. Reposition the initial keyword within the cocoon: 
   - As a **Daughter Page** (consideration) if it is a major axis. 
   - Or as a **Granddaughter Page** (informational) if it is a sub-axis. 
5. Then, build the cocoon based on the broadened keyword.

> Example: "Inclusive manager" (too narrow / mostly consideration) → Move up to "Management training" (transactional for a training provider). "Inclusive manager" becomes a Daughter or Granddaughter page depending on the logic.

---

### 1. Semantic Analysis of the Main Keyword (Transactional) 
- Dominant search intent (expected to be transactional) 
- Main and secondary entities 
- Variants, synonyms, long-tail keywords 
- SEO opportunities 
- Business performance criteria (conversion, lead, contact, quote request, etc.)

---

### 2. Architecture of the Semantic Cocoon in an Intent Pyramid 
**Mandatory Structure:**

#### A. Mother Page (Transactional) 
- **1 unique pillar page** - Positioned on the **broadest, business-driven transactional keyword** - Must be able to encompass the entire theme

#### B. Daughter Pages (Consideration) 
- **3 to 7 pages** - Each corresponds to a **major sub-theme** of the Mother subject 
- Targets queries such as: 
  - "Best solution for…" 
  - "Training X vs Y" 
  - "Price / Duration / Modalities / Certification" 
  - "For whom?", "How to choose?" 
  - "Pros / Cons", "Feedback/Reviews"

#### C. Granddaughter Pages (Informational) 
- **3 to 10 pages per Daughter Page** - Targets a discovery / learning intent: 
  - Definitions 
  - Practical guides 
  - Methods 
  - Mistakes to avoid 
  - Steps / Checklists 
  - Concrete examples 
  - FAQ / Long-tail questions

---

### 3. Main Keyword for Each Page 
For every page in the cocoon, you must provide: 
- **Page Title** - **Associated Main Keyword** - **Intent Type (Transactional / Consideration / Informational)**

⚠️ Important: No Hn headers, meta descriptions, or detailed briefs.

---

### 4. Optimized Internal Linking (Adapted to the Pyramid) 
**Rules:** - **Downward Linking**: Mother Page → Daughter Pages → Granddaughter Pages 
- **Upward Linking**: Granddaughter Pages → Daughter Pages → Mother Page 
- **Horizontal Linking**: Only between pages of the **same level and within the same silo** - **Anchor Text**: 
  - 60% Long-tail 
  - 30% Close semantic anchors 
  - 10% Exact match 
- **No cross-silo links** unless there is a strong semantic justification.

---

### 5. Editorial Schedule Proposal (Based on Intent) 
- Prioritize **Transactional and Consideration** pages first (direct business impact) 
- Follow with Informational pages to establish authority 
- Identify quick wins (low competition / clear intent) 
- Propose a logical and progressive publication order.

---

## Analysis Framework 

### 0. Intent Level Validation 
- Can the keyword be **transactional**? 
- If not: broaden towards a transactional mother theme.

### 1. Simulated SERP Analysis 
- Dominant page types (sales / consideration / info) 
- Recurring angles and entities 
- E-E-A-T differentiation opportunities

### 2. Thematic Segmentation 
- Coherent silos without overlap 
- Each silo = a **Consideration Daughter Page** - Each sub-topic = **Informational Granddaughter Pages**

### 3. Pyramid Hierarchy 
- 1 Transactional (Mother) 
- 3–7 Consideration (Daughters) 
- 3–10 Info per Daughter (Granddaughters)

### 4. Internal Linking 
- Downward / Upward / Horizontal intra-silo only 
- No cross-silo without justification

### 5. Schedule 
- Transactional → Consideration → Informational 
- Publication sequence oriented towards SEO results + Business impact

Tools to enable :

• ☑️ Retrieve SEO recommendations
• ☑️ Search on Google
• ☑️ Get the content of a web page

Click Create to save your agent.

2) Launch the analysis on a piece of content

Once the agent is created:

1. Open an existing piece of content in the Semji editor (or create a new one)
2. Make sure the Primary Keyword corresponds to the pillar keyword you want to analyze
3. In the sidebar, go to the AI+ Agents section
4. select your Semantic Cocoon agent
5. Click Launch

The agent will analyze your keyword, perform Google searches to validate its potential, and generate the complete architecture.

Tip ✅: For best results, choose a keyword that’s broad enough and aligned with your commercial offering. For example, prefer "management training" to "inclusive manager training".

3) Interpret the result

The agent generates a structured report including:

Keyword validation

• If your initial keyword is validated as a transactional parent page, the agent builds the cocoon directly
• If the keyword is too narrow, the agent proposes a broader scope and repositions your original keyword within the architecture

Cocoon architecture

• Complete list of pages to create, organized by hierarchical level
• For each page: title, main keyword and intent type

Linking strategy

• Linking recommendations between pages
• Optimized anchor suggestions

Editorial schedule

• Recommended publication order
• Identification of quick wins

Options and variations

To customize or extend the use of this agent:

• Adapt the industry sector: Modify the prompt to include specifics of your industry (e-commerce, B2B, services, etc.)
• Adjust the number of pages: Request a specific number of daughter pages according to your editorial production capacity
• Refine after generation: Once the cocoon is generated, ask follow-up questions such as "Can you add 3 additional granddaughter pages for the [X] silo?"
• Combine with other agents: Use the Content Gap agent to identify missing opportunities, then use the Semantic Cocoon agent to structure your response

Why is it useful?

How can you verify that it worked?

After execution, you can validate the quality of the result by checking that:

• The agent correctly identified or proposed a transactional parent page aligned with your offering
• Each daughter page corresponds to a distinct and relevant consideration axis
• The granddaughter pages cover precise and varied informational questions
• The proposed linking respects the silo logic (no unjustified cross-silo links)
• The editorial schedule prioritizes content with high business impact

Attention ⚠️: If your initial keyword was broadened by the agent, check that the new proposed keyword still matches your commercial positioning and your professional expertise. You can adjust it manually if necessary.

To go further

Refinement questions to ask the agent:

• "Can you detail the recommended anchor texts for linking between the mother page and the daughter pages?"
• "Propose 5 additional granddaughter pages for the silo [name of the silo]"
• What are the quick wins to prioritize for this cocoon?

Useful links:

• Create and configure a custom AI+ Agent
• Identify missing topics with the Content Gap agent
• Optimize the internal linking of your content