> For the complete documentation index, see [llms.txt](https://oodlz-ai-studio.gitbook.io/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://oodlz-ai-studio.gitbook.io/docs/outbound-agents/awareness/2-articles.md).

# 2. Articles

From an approved topic to a research-backed, SEO-optimised, image-rich website draft — autonomous, in one run, under $1 per post.

> **One approved topic in. One draft-ready website article out. No manual research, no manual writing, no manual SEO.**

Most AI tools take a keyword and return a blog post in 30 seconds. The Oodlz Articles producer runs a different problem: research the first page of Google for the topic, generate brand-native images and a thumbnail, write 2,500–3,500 words against a strict editorial brief with anti-AI rules embedded at prompt level, inject 3 internal links and external citations through a deterministic link pipeline, format the output as native CMS blocks with the full design system applied, and publish as a draft with SEO metadata pre-populated.

## What You Get

* A complete website draft per approved topic — title, slug, content, featured image, in-blog images, alt text, captions, meta title, meta description, focus keyword.
* Research brief reverse-engineered from the actual top-ranking competitors — search intent, content gaps, exploitable data, mandatory sections.
* 2,500–3,500 word article with 1–1.5% focus keyword density and anti-AI writing rules enforced at prompt level.
* 3 internal links injected per article via 7 deterministic validation checks — first valid candidate wins, no vague anchors, no link clustering.
* External citations as superscript footnotes with live URL validation before publish.
* Native CMS block formatting — branded H2 labels, styled pull-quotes, auto-TOC, FAQ accordions.
* SEO meta fields written via API — the post arrives with a green SEO score before review.

## How It Works

Each approved topic kicks off a sequential pipeline. The Research Agent makes 10–15 live tool calls — Tavily deep search, DataForSEO SERP scraping, full-page browsing of the top 5–7 ranking competitors — and produces a strategic content brief. The brief feeds the Image Generation Agent, which creates 3–4 unique image concepts with editorial metadata, and the Thumbnail Agent, which produces a dedicated featured image engineered for both hero and 200×200 thumbnail use.

The Writer Agent receives the full brief plus image metadata and produces the article in one structured pass — focus keyword density enforced, anti-AI banned phrases blocked, branded H2 labels applied, pull-quotes placed at natural pause points, FAQ sections structured for rich-snippet eligibility. A separate Meta Agent generates SEO meta title, meta description, and URL slug. A SANITIZE Citations step strips any non-approved source attributions deterministically before the link pipeline runs.

Internal and external links inject through the dedicated link architecture. 3 internal-link slots fire in article-position order so SEO first-link juice always flows to the most important destination. External citations validate URLs with live HEAD requests before injection. POST-PROCESS enforces final structural rules: 1 internal link in pre-H2 zone, 1 internal link per H2 across the body, no citations in intro or conclusion. Block formatting converts the result into native CMS blocks with the brand's design system applied. The post lands as a website draft with category, featured image, and SEO fields ready.

## Quality Baked In

* Real first-page-of-Google research — every article competes against the actual ranking content, not training data.
* 1–1.5% focus keyword density and anti-AI writing rules embedded at prompt level — Google's helpful content system has a harder time penalising the output.
* 7 deterministic link validation checks before any internal link is injected — no vague anchors, no anchors in headings or FAQ accordions, no clustered links.
* External URLs validated by live HEAD request — broken links never reach publish.
* Article-position-ordered link injection — first-link SEO juice always flows to the correct destination.
* POST-PROCESS final pass enforces structural rules deterministically — no LLM judgement, no surprises.
* Native CMS blocks with brand design system applied — pull-quotes, TOC, FAQ accordions all styled consistently.
* Sources block auto-renumbered in document order on every run — citations always sequenced correctly.

## Inputs

* One approved topic record from the Topics producer's [approval queue](/docs/outbound-agents/awareness/approval-gates/gate-1-topics.md).
* Brand profile fields — name, tone, prohibited language, internal link library, brand colours, audience portrait.
* Live search and research API access (Tavily, DataForSEO, Firecrawl).

## Outputs

* Website draft post with full content, featured image, in-blog images, slug, meta title, meta description.
* Article record in the workspace updated with CMS Post ID, CMS URL, and Pending Review status — ready for [approval](/docs/outbound-agents/awareness/approval-gates/gate-2-articles.md).
* In-blog images and thumbnail uploaded to the CMS media library with SEO metadata and folder organisation.

## Cadence And Volume

Runs per approved topic. One topic produces one draft. Cost per article under $1. Wall-clock time per run is multiple minutes — research is the longest stage and intentionally not rushed.

## What This Replaces

* The content team that manually researches the SERP, writes the brief, drafts the article, sources images, formats for the CMS, and adds SEO metadata — half a day per post.
* The AI writer that produces a 2,500-word article with no real research, hallucinated citations, and stock-photo aesthetics.
* The SEO plugin that asks the human to fill in the meta title and description after the post is already written.
* The link tool that injects vague anchors in headings and FAQ accordions and breaks the SEO architecture.

One pipeline, real research, real images, real links, real SEO — every article, autonomously.

![Outbound articles flow — sequential pipeline from approved topic through research, image generation, writing, meta, link injection, and block formatting to a website draft](https://2935174940-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9WhvO5Y38RJO9oNIdz9G%2Fuploads%2Fgit-blob-b6652b3a1fd3a0a8fad0c152137f1531caa22fb2%2Fdiagram-outbound-articles-flow.svg?alt=media)

***

*This was built. No agent ships this out of the box.*