Within our team, I am responsible for implementing websites and AI-powered workflows. I also examine how they can be integrated into existing processes and automations in a privacy-compliant way.
My work has been decisively shaped by 12 years in search engine optimisation (SEO).
That is why I follow clear rules in our projects: define your target audiences. Meet your customers’ search intent. Prioritise only the improvements that support your business goals. Create a clear user journey. And base decisions on data instead of instinct.
Today at 65 Creative
My part in the team.
I see my strength in looking at projects as a whole and examining the crucial details a second or third time. Care, thorough testing and an awareness of possible interactions – for example around data privacy – therefore land with me. My co-founder Konstantin brings the creative focus and speed needed to sharpen ideas and scale them quickly. These different strengths complement each other within our team.
SEO since 2014
Experience that shapes my perspective.
Over the years, I have supported more than 100 clients through projects, consulting and ongoing work, and closely examined over 1,000 websites.
Most of these clients came from healthcare, particularly plastic surgery.
That placed many of these websites within Google’s YMYL and E-E-A-T area. This includes topics where inaccurate information can influence decisions and create risks to health, safety or financial stability. Clear signals of quality and trustworthiness are therefore especially important.
Even if SEO plays a smaller role in the AI era, the principles learned through it remain important.
Today, anyone can create an appealing prototype quickly. Whether it can become a viable business idea only becomes clear through testing: can crawlers reach the content? Is the page accessible? Do external services load only after the required privacy consent?
That is where my testing and SEO background come in.
More about my SEO experience
Understanding ranking drops.
Even back then, companies came to me after their websites suddenly lost rankings. They often felt they were following Google’s recommendations but still could not explain the drop.
I understand why: Google’s information is spread across numerous documents, publications and individual statements. Some seem contradictory at first or only make sense in context. Running a business rarely leaves time to keep track of all of it and apply it to your own website.
This remains one of my specialisms: I combine on-page SEO and keyword research with a close study of Google’s quality and spam policies. This includes the Search Quality Rater Guidelines and the YMYL and E-E-A-T concepts, which address sensitive topics and content quality and trustworthiness.
My years of experience help me put changes in context. Some issues that stand out during a core update are ones I encountered years earlier. That gives me clues about where a closer investigation could be worthwhile.
From our daily work
A few AI examples. More time for what matters.
I remember the good old days of SEO well: after ten hours of keyword research, filtered Excel lists and endless backlink reviews, dry and tired eyes were inevitable.
A few examples from our daily work.
01
15 minutes for keyword research instead of 10 hours.
The everyday problem
Thorough keyword research used to take ten hours of manual work. Collecting and cleaning up keywords was followed by examining the content behind the search results.
What we built
Our workflow filters search terms and retrieves the top three to four results for each keyword. We then examine their content to assess search intent.
What this makes easier
We filter large keyword lists in very little time. Instead of losing focus in informational keywords, we identify the terms with purchase intent that matter to the business.
A look at the implementation
Search results come from the SE Ranking API. The content of those pages is then retrieved and analysed.
SE Ranking API
02
Website audits straight from an AI chat.
The everyday problem
After a technical website audit, results often need to be exported and prepared for further analysis.
What we built
I set up Screaming Frog on a separate server and connected it to AI applications. We can start website crawls directly from ChatGPT and process their output.
What this makes easier
Fewer manual steps between assessing a website and working on the problems we find.
A look at the implementation
An MCP interface makes Screaming Frog Spider on our VPS accessible to the AI chat. Crawl output can feed into subsequent automated tasks.
Screaming Frog · MCP
03
Knowing when an article needs attention.
The everyday problem
A good article can become outdated as its subject develops. Regularly rechecking every source is time-consuming.
What we built
Automated workflows use a client profile to research relevant new sources, for example every three months, taking account of the subject’s requirements.
What this makes easier
A specific alert instead of starting research from scratch: which article needs attention, and why?
A look at the implementation
For an inheritance-law article, this might be a new Federal Court of Justice ruling or a change in legislation. The client profile records requirements such as factual accuracy and the treatment of sensitive YMYL content.
AI · YMYL · E-E-A-T
04
Editing your own website through chat.
The everyday problem
Moving a client website from WordPress to Astro removed the familiar editor for around 150 pieces of content, including about 120 articles.
What we built
We built an AI chat as a new way to manage the site. The client describes the change. Editorial corrections can pass through the designated process without additional agency approval.
What this makes easier
The client can handle many tasks independently. Changes to a contact form or the function of an important button are sent to us for technical review.
A look at the implementation
The website runs on Astro and changes are processed through GitHub. The workflow distinguishes editorial changes from functional changes requiring our approval.
Astro · GitHub
05
Newsletters and website data in the same chat.
The everyday problem
Managing a website, newsletters and visitor analytics means learning and switching between several applications.
What we built
The same client chat has access to newsletter management and website analytics, allowing the client to retrieve information, record ideas and initiate connected tasks.
What this makes easier
One shared point of access makes everyday work easier and reduces the need to contact our agency.
A look at the implementation
Brevo provides newsletter management and PostHog provides website analytics. Both are connected to the client chat through interfaces.
Brevo · PostHog
06
An AI assistant for our agency’s daily work.
In use & expanding
The everyday problem
Transferring time records, compiling costs and tracking contract deadlines: even small administrative tasks add up.
What we built
Our employee’s sign-in and sign-out in Slack already transfers time data to our time-tracking system automatically. We are expanding this approach into an AI assistant for agency operations.
What this makes easier
Manual time-data transfer is already eliminated. Cost and contract overviews, plus early reminders about cancellation deadlines and other important dates, are in development.
A look at the implementation
The assistant is being developed using Hermes and n8n, with Slack as a shared point of access. It is intended to compile information for business decisions and handle recurring administrative work.
Slack · Hermes · n8n
Milestones
My professional journey.
2026
Co-founder & CTO
65 Creative · since 2026
2022
Freelance web developer
Web development · since 2022
2020–2023
Head of SEO
FirmCatalyst
2018–2020
SEO specialist
FirmCatalyst
2016–2018
Vocational training in digital & print media design
Vocational training
Sharing what I learn.
I have learned a lot from people who share their knowledge online. That is why I have published guides of my own, including articles for Raidboxes.
