Publish Google News–based SEO blog posts to WordPress with Claude, Gemini and RankMath

Stop writing blog posts manually. This workflow monitors Google News every 12 hours on any topic you choose, automatically selects the most relevant article, scrapes the source content, generates a 100% original SEO-optimized blog post using Claude AI, creates a featured image with Google Gemini, and publishes the complete post to WordPress with RankMath meta title and meta description fully automated, zero manual work. Perfect for bloggers, content marketers, digital agencies, WordPress site owners, and anyone who wants to keep their website updated with fresh, AI-written content without spending hours writing or publishing. --- What problem does this solve? Publishing fresh, well-written blog content consistently is one of the hardest parts of running a website. Finding relevant news, writing original articles, creating images, and publishing everything to WordPress takes hours every single time. This workflow eliminates all of that. It runs on a schedule, picks the best news story on your topic, writes a completely original article from scratch, and publishes it to your WordPress site, including featured image, meta title, and meta description while you focus on everything else. --- How it works Scheduled trigger runs every 12 hours automatically No manual action required. The workflow runs on its own twice a day. Fetch latest news via SerpAPI Google News SerpAPI fetches the freshest articles from Google News based on your search query. The default query is seo but you can change it to any topic - tech, finance, health, marketing, AI, real estate, or anything your blog covers. One parameter change is all it takes. Check already-published articles via Google Sheets A Google Sheets node reads all previously published news links from your tracking sheet. This ensures the same article is never processed twice - even if it keeps appearing in Google News days later. Remove duplicates with Merge node The Merge node compares incoming articles against your published history and removes any matches. Only genuinely new, unprocessed articles move forward. Sort by date and limit to top 10 Remaining articles are sorted by publish date, newest first. A Limit node keeps only the top 10 most recent items for evaluation. AI agent selects the single best article All 10 article titles and links are combined into one item and sent to an AI Agent powered by Claude (via OpenRouter). The agent reads every title and picks the one most relevant and valuable for your website's audience - filtering out off-topic, low-quality, or irrelevant results. It returns the winning article's title and link. Scrape the full source article An HTTP Request node fetches the complete HTML of the selected article. An HTML Extract node pulls only the meaningful content - headings (h1, h2, h3) and paragraphs (p) - stripping out ads, navigation menus, and everything else. An Aggregate node joins all extracted text into one clean block ready for the AI. Claude generates a 100% original article A second AI Agent (Claude via OpenRouter) receives the scraped content as research material only. It does not rewrite or paraphrase the source. It writes a completely new, original, SEO-optimized article based on the topic and ideas - with its own structure, wording, and insights. The output includes: - Article title - Meta title (under 60 characters) - Meta description (under 160 characters) - Full article body in Markdown format - A featured image generation prompt Google Gemini generates the featured image The image prompt produced by Claude is sent to Google Gemini's image generation model. The image is created in 16:9 ratio, suitable as a WordPress featured image. Upload image to WordPress Media Library An HTTP Request node uploads the generated image binary directly to your WordPress site via the REST API. WordPress returns an image ID which is stored for the next step. Convert Markdown to HTML The article content is in Markdown format. A Markdown node converts it to clean HTM