THE AI SEO OPERATING SYSTEM
Most teams use AI tools.
I built the system.
I turned the full SEO workflow—research, briefs, drafting, QA, publishing, localization, migration and maintenance—into one governed library of production agents and automations.
THE DISTINCTION
Not isolated prompts. One connected system.
Each module is a real, versioned workflow with its own rules, quality rubric and documented failure modes. The library runs in production for a seven-market European travel brand. Scheduled modules can operate autonomously; editorial judgment and publishing remain human responsibilities.
The system has grown beyond the 16-workflow subset documented in the roadsurfer case study. The figures here describe the current operating library; that case study preserves the earlier measured scope.
THE FIVE LAYERS
Content flows through one standard.
Demand and audience signals decide what deserves to exist: trend scouting, opportunity scoring, content-gap checks, keyword strategy and refresh triage.
Verified research becomes a demand-grounded brief, draft, fact-check, SEO/AEO optimization and humanized editorial review.
Publication QA, CMS formatting, taxonomy, structured data, internal links and a complete Gutenberg draft—publishing itself stays human.
Each market receives a native-demand verdict: adapt, refresh, re-angle or skip. Full adaptations begin with target-language keyword research.
Migration agents, read-only audits, restore procedures, health checks and refresh triage keep the site from quietly decaying.
Every layer reads the same editorial standard. Change the central rubric once, and the connected workflows inherit the new expectation—so page one and page one thousand can sound like the same brand.
DESIGN PRINCIPLES
Governance is part of the product.
- 01 / Skills, not prompts
Every workflow is versioned and documented with invocation rules, a quality rubric and known failure modes.
- 02 / Humans own judgment
Agents research, draft, format and build. People approve publication, live edits and destructive actions.
- 03 / One shared standard
A central editorial standard defines voice, SEO and AEO rules for every connected workflow.
- 04 / Right tool per job
Demand APIs, browser agents, automation, deterministic scripts and language models each do the work they suit best.
WHAT CHANGES
Automation moves my time toward judgment.
Research through publication-ready draft accelerates; my time moves to editing and decisions.
Shared rubrics replace memory and scattered checklists as the source of quality.
Adaptation begins with native keyword demand, not with forwarding a translation.
Legacy migration and read-only audits run with snapshots and restore procedures.
THE OPERATING MODEL
SEO where humans do the work that needs a human.
Agents handle repeatable research, drafting, formatting, building and checks. People own strategy, exceptions, quality decisions and publication. Supervised modes remain available even where narrow autonomous workflows have proven safe.
The stack combines agent skills, n8n, demand data, WordPress/Gutenberg, Google Search Console, GA4 and deterministic scripts. All workflows operate under company data-governance rules.
Discuss an SEO role ↗