AI SEO Automation Demo — 2026-04-01

Overview

During the ^[1], Mark live-demonstrated using a Claude/Cloud Code AI agent to automate WordPress SEO tasks on the ^[2] staging site. The workflow replaced hours of manual page-by-page work with a single agent run, completing meta data and alt text across the entire site in minutes.

The demo also surfaced a retainer proposal opportunity: Seamless is wrapping up their site build and has expressed interest in ongoing marketing spend.

The Workflow

Prerequisites

WordPress application password generated under the target user account (Users → Edit → Application Passwords)
Credentials stored in the master credentials spreadsheet so the AI agent can locate them automatically on live sites
Note: staging environments require a one-off password setup; it's easier to run this workflow post-launch when credentials are already in the master sheet

Step 1 — API Access

The agent used the WordPress REST API, authenticated via the application password, to enumerate all pages and media on the site.

Step 2 — Meta Data

The agent:
1. Fetched each page's HTML content
2. Identified target keywords from the page copy
3. Generated SEO-optimized meta titles and meta descriptions (respecting standard length limits)
4. Wrote the values back to the site via API (through the active SEO plugin)

Step 3 — Reporting

The agent:
- Generated a detailed Markdown report summarizing all changes made
- Converted the report to PDF
- Posted the PDF to Slack

Step 4 — Alt Text

The agent:
- Scanned the WordPress media library
- Generated descriptive alt text for each image based on filename and context
- Updated the media library records via API

Key Observations

Speed: Tasks that would take a human several hours of page-by-page editing were completed in approximately one minute of agent runtime.

Quality: The agent correctly inferred target keywords from page content and produced appropriately sized meta descriptions without manual prompting.

Supervision required: The workflow requires expert setup (API access, credential management, prompt engineering) and human review of outputs. It is not a fully autonomous, fire-and-forget process.

Staging caveat: Running on a staging site required extra credential setup steps. The recommended practice is to run this workflow on the day a site goes live, when credentials are already in the master spreadsheet.

Memory / documentation: After resolving a related issue (the ^[3] iframe/security-header fix), Mark instructed the agent to write memory files documenting what was found and where. The same pattern applies here — agent-generated documentation reduces future diagnostic time.

Proposal Context

Seamless is nearing launch and has budget to spend on ongoing marketing. Discussed approach:

Option	Details
Retainer (anchor)	$3,500/mo — Google Ads + SEO/content + meta/ongoing optimization
Retainer (floor)	$2,500–$3,000/mo if client pushes back
Performance hybrid	Lower base (e.g., $1,500/mo) + percentage of closed deals — positioned as "if we're successful, it costs you more"

Services in scope: Google Ads, Meta advertising, SEO, content, site performance optimization.

Next step: Mark to draft the formal proposal; Melissa to pitch at the next Seamless meeting (not the immediately upcoming week as of 2026-04-01).

Action Items

[ ] Mark — Draft Seamless proposal (Google Ads + SEO/content); anchor at $3,500/mo with performance-based fallback
[ ] Melissa — Present proposal at next Seamless client meeting

^[4]
^[5]
^[6] (candidate article — generalize this pattern)
^[7] (related agent + memory file pattern)