Autonomous AI SEO Agent for Sabai Health

We built an autonomous SEO content pipeline for Sabai Health, a Thai-language digital health platform. The pipeline runs entirely through Telegram. The operator types one command. The system handles keyword research, brief generation, article writing, medical review routing, and publishing to the live website. No CMS is ever logged into manually.

Sabai Health publishes Thai healthcare content. The work must meet Google's E-E-A-T standards. Manual writing is slow. Generic AI writing fails medical accuracy. For Thai-language medical content, there's no off-the-shelf solution.

•⁠ ⁠A Telegram bot as the only operator interface •⁠ ⁠9 n8n workflows orchestrated on a self-hosted VPS, every failure routed to a central error handler •⁠ ⁠Claude Sonnet 4.6 doing three jobs: keyword clustering, JSON brief generation, and full bilingual article writing with 5 quality gates •⁠ ⁠Semrush for live Thai keyword research •⁠ ⁠Google Sheets as persistent state and memory across every workflow •⁠ ⁠Notion as the medical review queue, with a polling gate that cannot be bypassed •⁠ ⁠Sanity CMS publishing directly to a Next.js frontend •⁠ ⁠Native Thai or English output, set once at /start

•⁠ ⁠1,998 production executions •⁠ ⁠0 failed executions, 0% failure rate •⁠ ⁠68 articles published in 15 days •⁠ ⁠9 workflows, 5 API integrations, 1.24s average run time •⁠ ⁠Every article medically reviewed before going live

The operator never opens a content tool. The system handles everything else. Healthcare content cannot be fully automated, so the pipeline doesn't try. It automates everything except the medical review, which stays human by design.

Stack: n8n · Telegram · Claude Sonnet 4.6 · Semrush · Google Sheets · Notion · Sanity · Next.js

Services: AI Agent Development · Workflow Automation · Telegram Bot · n8n Orchestration · Prompt Engineering