How I Use AI Agents to Run a More Structured SEO Workflow
Davis Christenhuis
When I began my SEO journey five years ago, I had to do almost everything manually. Keyword research, SERP analysis, content briefs, reporting, internal linking, competitor checks. It all took a lot of time, and most of it lived across different tools.
Now, working at Dust, my SEO workflow has changed a lot. I use AI much more in my day-to-day work, but I'm careful not to replace the parts of SEO that matter.
AI helps me move faster, improve the way I work, and spend more time on strategy, judgment, positioning, and understanding what the audience actually needs. In this article, I want to share four SEO agents that I use in my own workflow.
Agent 1: Finding new keyword opportunities
The first agent is SEOMonitor. This is a keyword research agent that helps us find new opportunities around a specific topic.
You can set it up around one keyword or a group of keywords. In our case, we use it for topics like “AI agent,” since that is one of the areas we want to keep a close eye on.
The setup is simple. The agent is connected to a Google Sheet where we keep the keywords we already monitor. When it runs, it pulls Semrush data for the topic, checks it against the sheet, filters for new keyword opportunities, and adds them back to the sheet with the useful details we need to evaluate them.
For each new keyword, it includes:
- Monthly search volume
- Keyword difficulty
- The top-ranking pages in the SERP
- Potential blog post title ideas based on the SERP
- Date detected
Because Semrush refreshes its keyword and SERP data over time, the agent can surface opportunities we may not have seen before. We let it run every week with a Trigger, so by the time we open the sheet, we already have the context we need to decide which keywords are worth exploring.
This is an example of what it can look like in your sheet.
Agent 2: Understanding the SERP before writing content
The second agent is SERPScout. This is the follow-up agent we use after finding a keyword that looks relevant.
The idea is simple. If the SEOMonitor agent finds a keyword opportunity, we still need to understand if it is actually worth writing about. Search volume and keyword difficulty are useful, but they do not tell the full story. We also need to understand what is ranking, who is ranking, and whether the keyword actually fits Dust.
For this agent, we only need to provide the keyword. From there, it gives us a more detailed overview of the SERP.
The first thing it does is a keyword suitability check. It checks internal knowledge, including our Notion pages, content guidelines, and content planning, to see if the keyword is actually relevant for Dust.
After that, it looks at the top 10 SERP results. For each result, it gives us:
- The page title
- The URL
- A short summary of what the page covers
- The headers used in the article
- The publisher type
Then it gives a summary of the SERP as a whole. This includes the most common headers, the publisher breakdown, and whether competitors are writing about the topic too.
That helps us understand what kind of content is currently ranking and what format search engines seem to prefer.
The final output is a proposed content structure, which is useful as a starting point. It shows which sections usually appear in the SERP, where there might be gaps, and how the article could be structured if we decide to write it.
This is what the SERP summary can look like.
Agent 3: Turning research into a first draft
The third agent is SEOArticles. This agent takes the research from SERPScout and turns it into a first draft.
If we find a relevant keyword and the SERP analysis shows that it is worth covering for Dust, we can use this agent to create a first draft. Once we decide on the structure, the agent turns it into a full SEO article with the important SEO/GEO elements already included. Depending on the article and the research needed, it can take around five minutes to write the full draft.
The output includes:
- Meta title
- Meta description
- Slug
- TL;DR
- H2s and H3s
- Full article draft
- FAQ section
- CTA suggestions
- Internal linking suggestions
- Sources
The draft is usually around 2,000 to 2,500 words.
The instructions for this agent are quite specific. They include:
- Tone-of-voice guidance
- Good and bad writing examples
- Rules for how to describe Dust
- Hard rules about what not to claim
- Structure requirements for SEO
- Guidance for LLM and GEO/AEO optimization
- Instructions to cite or list sources
- Instructions to avoid making things up
So instead of starting from a blank page, we start from a structured draft that already includes the main SEO elements. The work becomes editing, fact-checking, and making the article sharper.
Interested in an example or the instructions? Send an email to [email protected].
Agent 4: Turning SEO performance into a Frame
The fourth agent is SEOVisualizer. This agent turns SEO performance data into a Frame, which is an interactive dashboard created by a Dust agent. Instead of sharing raw data or a static report, the Frame gives people one place where they can explore the results.
As an SEO specialist, it can sometimes be difficult to explain what you are working on and what results it is driving. The data usually does not live in one place, which makes it harder to quickly show what is working, what is not working, and where the opportunities are.
The Frame shows metrics and insights such as:
- Total sessions
- Signups
- Form submissions
- Articles that are performing well
- High-priority opportunities
- Medium-priority opportunities
- Low-performing pages
- Key takeaways from the content performance
SEO reporting is not only useful for SEO people. It helps other teams understand which articles are performing well, which ones are growing, and which pages might need more work. It also makes internal analysis easier, because we can quickly see what to optimize next.
For this example, the Frame uses demo data instead of real performance data.
What changed in my SEO workflow
These agents help me keep up without adding more manual work. They make parts of the workflow easier to repeat, improve, and share with the team.
That is probably the biggest change for me. I spend less time rebuilding the same process from scratch and more time thinking about what actually matters: strategy, quality, positioning, and where SEO can drive growth.
Instead of jumping between tools and starting over every time, I can use agents to make the workflow more structured while still keeping control over the final decisions.
This is where multiplayer AI becomes useful. The agents are not just prompts I use on my own. They are workflows that other people can reuse, improve, and build on.
To watch the agents in action, you can view the full webinar, where I walk through the setup and demos.
Explore how Dust can support your SEO workflow. Get in touch →