"AI-powered" has joined "data-driven" and "full-service" in the graveyard of marketing buzzwords that mean everything and nothing. Every agency claims AI capabilities. Most mean they use ChatGPT to write first drafts of blog posts.
We built Omakaase on a genuine AI-assisted workflow from the start. Here's an honest look at what we use, where AI actually helps, and where it still falls short.
What AI is genuinely good at in marketing
Research at scale
The part of SEO research that takes the most time is analysis — processing large amounts of data to find patterns. AI tools dramatically accelerate: competitor content gap analysis, SERP feature analysis across hundreds of keywords, schema validation across large site crawls, and audience research synthesis. Work that took a strategist a full day now takes two hours.
First drafts and structure
AI is excellent at producing structurally sound first drafts — clear headings, logical flow, reasonable coverage of a topic. The problem is that first drafts are generic. Good marketing content has a point of view, uses specific data, and reflects real experience. That still requires human editing.
Personalisation at scale
Our programmatic page system generates market-specific content for 2,000 service × industry × city combinations — each page unique to that market. Without AI, this would require thousands of hours of manual writing. With AI, it requires careful prompt engineering and human quality review.
Where AI consistently falls short
Strategy
AI can tell you what the average answer to a marketing question is. It cannot tell you what the right answer is for your specific business, competitive position, and customer base. Every AI-generated strategy recommendation needs a human with actual market knowledge to validate it.
Differentiation
AI models are trained on existing content. They produce outputs that look like an average of what already exists. Marketing that genuinely differentiates a brand requires insight that goes beyond what's already been published — which AI, by definition, cannot provide.
Relationship-driven content
Case studies, thought leadership, brand voice, client stories — these require human experience and judgment. AI can help structure and edit this content, but it cannot originate it.
Our actual stack
- Research: Perplexity for real-time research, Claude for synthesis and analysis
- Content: Claude for first drafts + structural outlines, human editors for final pass
- SEO analysis: Screaming Frog + custom scripts for technical audits
- Paid: Google's Performance Max + manual campaign builds (we don't trust fully autonomous bidding for most clients)
- Reporting: Looker Studio with custom GA4 + Search Console connectors
- Programmatic pages: Custom generation pipeline using Claude API with proprietary prompts
The honest conclusion
AI has made our research faster, our content production more scalable, and our technical analysis more thorough. It has not replaced the need for experienced strategists who understand marketing fundamentals and know how to apply them to specific business contexts.
Any agency telling you that AI has replaced the need for experienced marketers is either confused about what marketing is, or trying to charge agency rates for automated output. The value in marketing has always been judgment. AI tools make it easier to execute on good judgment — they don't generate judgment themselves.