Automate monthly refreshes of data-driven articles with an agent that fetches, filters, rebuilds and drafts for human review

Articles built on data (top searches, top questions and similar lists) go stale fast, and refreshing them by hand is slow, rule-heavy work that tends to slip from monthly to quarterly.

1. Pick one repeatable, rule-based job that keeps getting delayed; here, refreshing data-led posts. 2. Write down every step, including unwritten rules and exceptions. The author suggests letting an agent interview you one question at a time and saving the result as a reusable skill file. 3. Connect trusted data sources (first-party data and APIs) rather than relying on the model's memory or open web search: the model processes the data but is not the source of it. 4. Give the system accepted and rejected examples with reasons, and keep a log of rejected items so rules can be checked. 5. Build one dataset end to end with a visible output at each stage: fetch, clean, classify, build, update the CMS draft, notify. 6. Compare the output side by side with the human version (numbers, rejections, formatting, tone, links), then schedule runs after the source data refreshes, with failure alerts. 7. Add datasets one at a time; widen the agent's permissions slowly (the author waited two to three months before allowing larger edits).

The author says the hub now refreshes 12 datasets each month and shows screenshots of drafts waiting for review and of rejected-item lists (including real questions wrongly rejected). Teammates adopted similar workflows. No time saved, cost, or traffic change caused by the automation is reported.

Vendor content: Ahrefs sells the data tools used and the agent platform featured, so the setup is easiest for its own customers. Run by a marketing team at a large company, not a solo founder. No before-and-after numbers. The author lists real failure modes: model updates changing how rules are read, source format changes, rule drift, CMS formatting breaks, and an access error caused by a colleague's workspace change. Human review stays mandatory.

A repeatable data-driven content job, an API or export for the data, a CMS with drafts, and an agent tool that can connect to them.

Start with a single dataset and keep the agent limited to drafts until its output matches yours for several runs.

Interview me about how I currently update [article or report], one question at a time. Follow up on any vague answer. When done, write a step-by-step checklist that includes every rule and exception I mentioned.

• Ahrefs blog (Si Quan Ong) ↗ Vendor case study, published October 16, 2026

Source published: October 16, 2026. Last reviewed here: October 25, 2026. Spot a mistake? Tell us.