AI-Assisted Content Migration: From Discovery to Deployment

84EM.COM ↗ had “13 years of experience” and “30 years of programming” hardcoded across 60+ pages. Every January, I’d need to find and update these references. The task seemed straightforward until you start counting the variations: hero blocks, template files, generated local SEO pages, FAQ sections.

Rather than manually hunting through the WordPress admin, I handed the problem to Claude Code.

Discovery: Let AI Do the Counting

The first step was understanding the scope. Claude generated and ran a database query to find all instances:

SELECT post_type, COUNT(*) as count FROM wp_posts
WHERE post_status = 'publish'
AND (post_content LIKE '%13 year%' OR post_content LIKE '%30 year%')
GROUP BY post_type;

Result: 63 content items across pages, reusable blocks, and templates. More than expected. The query also surfaced variations I hadn’t considered – “nearly 30 years,” “over 13 years,” “13+ years.”

This is where AI shines. I described the problem; it figured out how to measure it.

Solution Design: AI as Architect

I explained the goal: replace hardcoded years with something that updates automatically. Claude proposed three approaches:

1. Content filter – Intercept content on render and replace text
2. Shortcodes – Store base years, calculate dynamically
3. Custom blocks – Gutenberg blocks with year logic

It recommended shortcodes as the simplest, most maintainable option. No magic behavior, explicit in the content, easy to debug. I agreed.

Claude then designed the shortcode API:

2026  → calculates years since 2012
14                 → alias for WordPress experience
31                → alias for programming experience

Clean, flexible, self-documenting.

Implementation: Code Generation

With the design approved, Claude wrote the implementation: a PHP file with the shortcodes and a WP-CLI command class for the migration.

The migration command included something I wouldn’t have thought to add: a dry-run mode.

wp 84em dynamic-years migrate --dry-run

This previews every change without modifying the database. Essential for a migration touching 63 pieces of content. Claude’s instinct for defensive programming saved potential headaches.

The Migration: 68 Replacements in Seconds

The dry-run showed 68 replacements across 62 items. Some pages had multiple year references. After reviewing the preview, the actual migration took seconds:

wp 84em dynamic-years migrate

Every instance updated consistently. No missed pages, no typos, no “13 years” on one page and “14 years” on another.

What AI Handled vs. What I Did

Claude Code handled:

• Database analysis and scope assessment
• Solution architecture and trade-off analysis
• Full PHP implementation with WordPress coding standards
• WP-CLI command with dry-run safety
• Case-insensitive pattern matching for variations
• Documentation

I handled:

• Describing the problem
• Choosing between proposed solutions
• Reviewing the dry-run output
• Running the final migration
• Verifying a few pages in the browser

The ratio was maybe 10 minutes of my attention for what would have been hours of manual work, and the result was more thorough than I would have achieved alone.

The Pattern

This workflow applies to any content migration:

1. Describe the problem in plain language
2. Let AI analyze the scope with database queries
3. Review proposed solutions and pick one
4. Get implementation with safety features built in
5. Dry-run first, then execute

The key insight: AI excels at the tedious parts (finding all instances, writing migration code, handling edge cases) while humans handle the judgment calls (is this the right approach? does the preview look correct?).