We manage 159 personal brand websites through BlitzAdmin. Each one runs WordPress with the same theme, same Rank Math SEO setup, and same Content Factory pipeline. The problem was that keeping all 159 sites healthy, publishing content on schedule, and catching SEO issues required a team of VAs doing the same repetitive checks over and over.
So I built an automation system using Claude AI agents that runs the entire MAA loop — Measure, Analyze, Act — across every site, every week, with zero manual intervention. Here is exactly how it works and why it matters.
The problem with managing 159 WordPress sites manually
Every personal brand site needs the same things: fresh content published on a staggered schedule, working pages that load fast, proper SEO configuration through Rank Math, a Topic Wheel with at least 9 pillar topics, and social proof through interviews and podcast episodes. Multiply that by 159 sites and you get a QA nightmare that no team of humans can maintain reliably.
Before automation, we relied on weekly spot checks. A VA would manually log into BlitzAdmin, click through a handful of sites, eyeball the content, and maybe catch the most obvious issues. Sites would go down for weeks without anyone noticing. Content would stop publishing because a draft got stuck. SEO settings would be misconfigured and nobody would know until organic traffic flatlined.
The architecture: WordPress REST API plus local credential caching
The first decision was to eliminate the browser entirely. Claude in Chrome is powerful for one-off tasks like building a roofing site from scratch in 45 minutes, but for fleet operations across 159 sites, you need API access. WordPress has a full REST API that supports every operation — reading posts, creating content, updating pages, checking settings, managing media — all through simple HTTP requests with Basic Auth.
BlitzAdmin stores the credentials for all 159 sites behind a dashboard that requires a JWT token. That token expires every 24 hours, which is fine for human use but terrible for automation. So we cache the entire site list with WordPress Application Passwords locally. WordPress Basic Auth credentials do not expire. This means the automation runs independently of BlitzAdmin’s token lifecycle. The only time a human needs to touch the system is when we add or remove a client site from the fleet.
Three scheduled tasks replace nine manual jobs
We consolidated what used to be nine separate weekly checks into three automated tasks:
Friday Full Fleet Audit (9 AM) — The agent scans all 159 sites every Friday morning. For each site it checks: is the site up or down, when was the last post published, how many total posts exist, are the core pages present (About, Services, Contact, Blog), is Rank Math configured with Person schema, and what does the SEO look like via Ahrefs (domain rating, organic keywords, backlinks). It classifies issues by priority — P0 for sites that are completely down, P1 for dormant sites with no recent content, P2 for missing pages or SEO gaps. The full report gets emailed automatically to web@blitzmetrics.com as a formatted HTML email.
Daily Content Factory Processor (10 AM weekdays) — This agent handles the Post stage of the Content Factory. It checks Google Drive for new transcripts and processed content, reviews draft posts sitting on client sites, and publishes them with proper date staggering. The staggering rules matter: podcast episodes go out on Mondays and Thursdays, interview articles every 3-4 days, and never two posts on the same day. This prevents the spammy look of bulk publishing while keeping a steady cadence that Google rewards.
Site Cache Refresh (manual, on demand) — When we onboard a new client or remove one, someone runs this task to pull the latest site list from BlitzAdmin and update the local credential cache. This is the only task that requires a human to be logged into BlitzAdmin.
How the Content Factory maps to automation
The four stages of the Content Factory are Produce, Process, Post, and Promote. The first stage — Produce — is the one thing that stays human. The client records a 2-3 minute video about their expertise using their phone. That is the raw input.
Everything after that is automated. Process: the AI agent takes the transcript, structures it into an SEO-optimized blog post following the client’s Topic Wheel, adds internal links to their pillar content, and formats it with proper headings and schema markup. Post: the agent publishes it to the right WordPress site via the REST API with the correct categories, tags, featured image, and Rank Math metadata. Promote: the content gets queued for dollar-a-day amplification across the client’s social channels.
This is the MAA loop running at scale. We Measure site health and content gaps through the Friday audit. We Analyze the data to prioritize which sites need attention. We Act by publishing content and fixing issues through the daily processor. Then we measure again the following Friday to see if our actions moved the needle.
What the first fleet audit revealed
The first automated audit across all 159 sites exposed problems nobody knew existed. 25 sites were completely DOWN — returning errors or unreachable. 104 sites were DORMANT, meaning they had not published a new post in over 30 days. 87 sites had zero posts at all. Only 24 sites were actively publishing content.
That is an 85 percent dormancy rate. Before automation, we had no visibility into this. The audit report now goes to the web team every Friday morning so they can triage the P0 issues (sites that are down) immediately, then work through the P1 and P2 items during the week.
The tools that make this work
The entire system runs on Claude Cowork with these integrations:
WordPress REST API — Every site operation (read, create, update posts and pages) goes through authenticated API calls. No browser, no clicking, no screenshots.
BlitzAdmin Dashboard API — Provides the master list of all sites and their credentials. Used only during cache refreshes.
Ahrefs MCP — Pulls domain rating, organic keyword counts, backlink profiles, and site audit data for SEO monitoring across the fleet.
Gmail SMTP — Sends the formatted audit reports directly to the web team via email. No manual forwarding or copy-pasting.
Google Drive — Source of truth for new content. Transcripts and processed articles land here from the Produce and Process stages, and the daily agent picks them up for publishing.
Cowork Scheduled Tasks — Cron-based scheduling that triggers each agent at the right time without any human needing to remember or click anything.
What this means for agencies running personal brand sites
If you are running more than 10 WordPress sites for clients, you cannot afford to manage them manually. The math does not work. A VA checking 10 sites per hour means 159 sites takes almost 16 hours of manual labor every week — just for basic health checks, not even content publishing.
With this automation, the Friday audit runs in about 30 minutes across all 159 sites and the report shows up in your inbox without you lifting a finger. The daily content processor handles publishing while you sleep. The only human input required is the client recording a short video on their phone.
That is the Content Factory working the way it was always designed to work — humans create the authentic, first-person content that proves E-E-A-T, and machines handle everything else at scale. The system does not replace the human element. It amplifies it across 159 sites simultaneously.
We will keep iterating on this. The 500 million tokens I burn every month are paying for themselves many times over when a single agent run replaces 16 hours of VA work. And unlike a VA, the agent runs the exact same checklist every single time without skipping steps or getting tired.