Query fan-out expert agent: identify all search intents around a keyword
Turn a single strategic keyword into an exhaustive semantic map and automatically produce the missing content sections to dominate both the SERP and AI engines.
At a glance
- What you'll do: set up an agent that explores all possible search intents around your main keyword, identifies angles not covered in your existing content and writes the missing sections in a format optimized for generative AI engines
- Time required: 10 to 15 minutes for setup, a few minutes per analysis depending on the semantic depth of the topic
- Ideal for: SEO leads and content managers looking to maximize the semantic coverage of their pillar content
What is it for?
Ranking well on your main keyword is no longer enough to capture all available traffic. Users phrase their searches in dozens of different ways, and generative AI engines like ChatGPT or Perplexity respond to highly varied prompts around the same topic. If your content only addresses one intent, it misses out on the vast majority of possible queries.
Query fan-out is the technique of starting from a single keyword and mapping all the probable phrasings a user might type or dictate to an AI. Each phrasing represents a distinct intent, and each uncovered intent is a missed traffic opportunity.
This agent automates that exploration. It uses Semji's SEO and GEO data to map the semantic space of your topic, compare that map against your existing content, and write the missing subsections directly in a question-and-answer format designed for featured snippets and AI-generated responses.
Good to know ℹ️: query fan-out is a concept from the world of AI search engines. When a user asks ChatGPT a question, the engine internally generates several variants of that question to cover all dimensions of the topic. Optimizing your content for fan-out means making sure you're relevant for each of those variants.
What you'll get
At the end of the analysis, your agent delivers a structured output in three parts:
- A complete fan-out map: 8 to 12 probable user prompts around your keyword, covering all intents (informational, comparative, transactional, situational)
- A coverage assessment: for each identified prompt, the agent indicates whether it is already addressed in your content, partially covered or entirely absent
- Written subsections ready to integrate: for each missing angle, the agent produces a question-and-answer paragraph optimized for featured snippets and AI citations, with Before/After blocks to make integration easy
[Suggested illustration: screenshot of the fan-out map generated by the agent with the color coding for covered/partial/missing]
Tip ✅: before launching the agent, make sure your content is open in the Semji editor and that the main keyword is filled in the "Focus Keyword" field. The agent relies on this data to calibrate its semantic exploration.
How to do it
1) Access the AI+ agents
From your Semji workspace, open the content you want to enrich in the editor. Then click the AI+ Agents icon in the right-hand sidebar to open the agents panel.
[Suggested illustration: screenshot of the Semji sidebar with the AI+ Agents icon highlighted]
2) Create the custom agent
Click Create an agent, then fill in the following fields:
- Name: Query fan-out expert
- Instructions: copy and paste the prompt below.
- Tools to activate:
- ✅ Retrieve SEO recommendations
- ✅ Retrieve GEO recommendations
- ✅ Get web page content
- ✅ Search on Google
[PROMPT]
Click Create to save your agent.
3) Run the analysis
Once the agent is created, select it from your agents list. The agent starts its analysis automatically from the content open in the editor. No additional instruction is needed: it retrieves the focus keyword, queries Semji's SEO and GEO recommendations, then launches its semantic exploration.
If you want to steer the analysis toward a specific angle, you can add a note in the conversation field before sending.
Example launch message:
Focus the analysis on comparative and situational intents, the topic is "air-to-water heat pump".
[Suggested illustration: screenshot of the conversation panel with the fan-out map being generated]
4) Interpret and integrate the results
The agent's report is presented in three successive blocks:
- Fan-out map: the complete list of identified intents and prompts, with their category (informational, comparative, transactional, situational)
- Coverage assessment: intents already addressed, partially covered and absent from your content
- Written subsections: content blocks ready to integrate, in question-and-answer format with Before/After blocks for each addition
To integrate the new sections, copy the generated blocks and paste them directly into the Semji editor at the location suggested by the agent. Then check your content score to measure the impact of the additions.
Using the agent day to day
This agent is particularly effective in two situations: when creating a new pillar content piece, to define an exhaustive semantic structure from the start, and when optimizing existing content that has plateaued despite a solid technical score.
Plan a fan-out analysis on your 5 to 10 strategic content pieces once or twice a year. Search intents evolve with trends and new user phrasings, especially as the use of conversational AI continues to grow.
Pro tip 💡: after integrating the new sections, re-run the ChatGPT content expert agent on the same content to check whether the additions actually improve your position in AI citations.
Go further
Once the fan-out map has been applied, you can deepen your semantic strategy by combining this agent with other Semji tools:
- Cross-reference with SEO intent analysis: make sure the newly added sections are consistent with the main intent detected by Semji to avoid sending contradictory signals to Google.
- Measure the impact on your content score: the added subsections should push your score forward. If they don't, ask the agent about the missing semantic entities in the generated blocks.
- Export the fan-out map as the foundation for your editorial calendar: each uncovered intent can become a standalone article if search volume justifies it.
Watch out ⚠️: the agent generates subsections that are ready to integrate, but they should be reviewed before publishing. Pay particular attention to figures and factual data sourced by the agent from external URLs, as these may change over time.
Useful links: