All postsAutopost already wrote and published articles for your brand on a schedule. It had one weak spot: someone had to keep feeding it topics. When the list ran out, the slots simply failed. Our own Life Future site hit exactly that, eleven empty slots in a row before anyone noticed. Today autopost learns to find its own topics, from what people actually search for.
Topics from real searches, not guesses
Pick Researched for SEO as the topic source and give Atlas a handful of seed keywords, such as the product areas you write about. Research then looks in four places:
- Google Search Console, when you connect it: the searches your site already shows up for, but on page one or two instead of the top. A focused article is often all it takes to move those up.
- Google search suggestions for each seed keyword, the phrases people type, like "ghost touch, do I have to replace the screen".
- Your most read articles, so the next post extends what already brings readers in.
- Trends, for seasonal and timely topics.
Anything that repeats what your brand already published is dropped before and after the AI looks at it. The AI then scores what is left and turns the best ideas into briefs.
A brief, not just a title
Each researched topic comes with the main search phrase to target, the reader's intent (looking for an answer, needing a service, ready to buy, comparing options), an outline, the questions the article must answer and related searches. The writer follows the brief: the keyword goes in the title, the meta description and the opening, the outline shapes the sections, and the questions become the FAQ.
It picks the right category
Every autopost article now lands in one of the categories you already set up. The AI chooses the one that fits each article, never invents a new one, and falls back to your default category when nothing fits. This applies to manual topics too, not only researched ones.
A quality gate before anything goes live
Publishing three AI articles a day without a check is a good way to hurt a site's ranking. So when auto publish is on, a researched article is checked first: the keyword in the title and meta description, enough sections, a real FAQ with at least three questions, no repeated paragraphs, and sensible title and description lengths. An article that falls short is not published. It waits as a draft with a review request for your publishers, and the reason is written down so you can see why.
It keeps itself fed
Research runs on its own every week, and sooner when fewer than five topics are waiting, at most once a day. You can open the research queue in the autopost settings to read each brief, skip a topic you do not want, queue it again later, or start research right away. Every research run counts against your monthly AI quota like any other AI call.
Connecting Search Console
Go to Settings, Integrations, SEO and click Connect. Atlas asks Google for read only access to Search Console, nothing else. After you pick your site's property, research starts reading the searches you already appear for. The access token is stored encrypted, never shown, and you can disconnect at any time.
Do I need to do anything?
Nothing changes until you choose Researched for SEO in the autopost settings; your manual topic list keeps working as before and stays saved if you switch. Connecting Search Console is optional, but it is the strongest signal research has, so we recommend it.
