Content & SEO Agent AI

Invoked when: weekly content calendar cycle runs — the next 2 articles scheduled in the calendar are briefed for the writing team

• –Pulls the target keyword for each scheduled article from the content calendar, then builds the full search intent picture for that keyword: the primary intent (informational, commercial, transactional, or navigational), the secondary intents that searchers with the same query also have (as indicated by the People Also Ask and related searches data), the top-ranking pages for the keyword and the content structure they use (word count, heading structure, content types — listicles, how-to guides, comparison pages), and the semantic keyword set that should be covered for the article to be topically comprehensive in Google's assessment. This is the strategic context for the brief — it tells the writer not just what the target keyword is but what the article needs to be to rank for it.
• –Structures the article brief with the following components: target keyword and search volume, secondary keywords to incorporate naturally, recommended article title (optimised for click-through rate as well as keyword inclusion), recommended meta description, content structure with H2 and H3 heading suggestions (drawn from the heading patterns in top-ranking pages and the People Also Ask data), word count target, content type recommendation (guide, listicle, comparison, or FAQ-heavy), mandatory topics to cover for topical completeness, and questions that the article should answer (which become the basis for the FAQ schema generated at the audit stage). The brief is a writer's working document, not an SEO data dump — it provides the strategic context in a format the writer can act on directly.
• –Identifies whether the scheduled article fills a gap in LendingIQ's existing content coverage (a topic not currently addressed in the published corpus) or improves an existing article that is ranking on page two and could reach page one with better on-page optimisation. Gap-filling and improvement briefs are structurally different: a new article brief focuses on comprehensive topical coverage; an improvement brief focuses on the specific optimisation changes (missing sections, thin coverage of secondary keywords, schema gap) that are holding the existing article back. The brief type is indicated clearly so the writer understands whether they are creating from scratch or optimising an existing piece.

Invoked when: an article is submitted to the CMS for editorial review — audit runs before the editor's review, so SEO issues are corrected before publication

• –Runs the 18-point on-page SEO checklist against every submitted article. The 18 checks cover: target keyword in the H1 (exact match or close variant), target keyword in the meta title (within the first 60 characters), target keyword in the meta description, target keyword in the first 100 words of the article, secondary keywords present in H2 headings, internal links to at least 3 relevant existing LendingIQ articles, external links to authoritative sources where factual claims are made, image alt text for all images (keyword-relevant where appropriate), article word count within 15% of the brief target, reading level appropriate for the target audience (Flesch-Kincaid Grade 10 or below for retail borrower content, Grade 12 or below for MSME content), no duplicate title tag with any existing published page, no canonical tag conflicts, URL slug containing the target keyword, mobile rendering check (no horizontal scroll, readable font size), page load speed indication (image file size check), structured heading hierarchy (H1 > H2 > H3, no skipped levels), FAQ section present (enabling FAQ schema), and no thin content sections (paragraphs under 50 words that address a heading topic without adequate coverage).
• –Produces an audit report with a pass/fail score for each of the 18 checks, a total audit score (number of checks passed out of 18), and specific correction instructions for each failed check. The correction instructions are actionable — not "improve meta description" but "add the keyword 'MSME loan eligibility' to the meta description within the first 100 characters, and ensure the description is between 120 and 155 characters." An audit report with vague feedback is not useful; each failed check needs a specific correction the writer can implement immediately.
• –Flags articles with audit scores below 70% (13 out of 18 checks passed) as requiring significant SEO revision before publication, and routes them back to the writer with the correction report rather than to the editor. Articles with scores of 70% or above proceed to editorial review with the audit report attached — the editor can see exactly which checks passed, which failed, and what the residual SEO issues are, and can make an informed decision about whether to publish with the remaining issues or request further revision.

Invoked when: an article passes the editorial review and is approved for publication — schema markup is generated and validated before the publish event

• –Generates Article schema (JSON-LD) for every published article: the article title, description, author (the human writer who produced the article — not the AI agent), publication date, modification date, image URL, and the LendingIQ organisation entity. Article schema is the baseline that confirms to Google that the page is a journalistic or editorial article, and it is required for eligibility in Google News and Discover surfaces where relevant. The agent generates the complete JSON-LD block; the CMS team inserts it into the page head before publication.
• –Generates FAQ schema (JSON-LD) for every article that contains a FAQ section — which is every article that was briefed by this agent, since FAQ sections are mandatory in the brief structure. The FAQ schema lists every question and answer in the article's FAQ section as a structured Question and AcceptedAnswer entity pair. FAQ schema is the primary mechanism through which LendingIQ's content can appear in Google's featured snippet FAQ accordions — a significant organic visibility opportunity for high-intent queries where LendingIQ has a comprehensive answer. The agent extracts the Q&A pairs directly from the article's FAQ section content; it does not generate new questions and answers independently of what the writer has written.
• –Validates every generated schema block against the Google Rich Results Test API before submission for publication. A schema block with a syntax error or a missing required field will not produce rich results and may produce a Search Console coverage error — validation before publication catches these issues before they are indexed. Where the validator returns an error, the agent corrects the schema and re-validates before delivering the final block. The CMS team receives only validated schema for publication; unvalidated schema is never passed for insertion.