Building Orena’s agentic content engine

Orena’s content workflow has published 148 articles and completed 45 refresh tasks. I built it to connect topic planning, drafting, review, publishing and search feedback in one process.

People researching memory and thinking often need help understanding cognitive assessments before they are ready to buy one. I wanted Orena’s articles to answer those questions clearly and build an organic acquisition channel over time.

A topic map connects broad guides to specific questions. The publication queue gives each article a purpose and records where it stands.

For each task, the agent loads the topic, related pages, editorial instructions and publication state. It prepares the article and assets, runs build and validation checks, and packages changes for review. The work lives with the website in version control, so I can inspect what changed and why.

The publishing process also handles metadata, internal links, page generation and the sitemap. Article content is present in the page’s HTML. These details are part of the workflow rather than tasks I have to remember at the end.

The refresh queue turns Search Console observations into specific tasks: clarify an answer, revisit a title, improve a link between pages or collect more data.

For example, an article explaining cognitive test scores was appearing in search but receiving relatively few clicks. Its refresh task focused on a title that better matched the reader’s question. The record kept the reason for the change and the wording before and after it.

That leaves a traceable sequence: a page, an observation, a change and a later check. Completing the task tells us what was done; subsequent search data tells us whether it helped.

Health content needs more than a passing build. An article can be technically correct in structure and still overstate a claim, misread a source or give a worried reader the wrong impression.

The editorial instructions call for plain language, credible sources and a clear distinction between education and diagnosis. Automated checks cover structure and technical issues. Factual accuracy, source relevance and the meaning a reader takes away still need human scrutiny.

The same approach guides search and answer-engine optimization: clear questions, direct explanations and accessible pages with useful supporting material.

Orena now has a content library and a record of what to publish or revisit next. I can pick up the work with its context intact, from the original topic decision to the latest refresh.

The 148 articles and 45 completed refresh tasks describe work delivered. Search impressions and clicks show how that work is being discovered. The longer-term test is whether it brings relevant visitors to Orena and helps them take a useful next step.