FAQ

Quick answers to the questions people ask most. Each answer links to the page with the full details.

Product

What does AgenticSEO do?

AI agents run a weekly loop for your site. They find the prompts and keywords your buyers use, write articles, check them with a QA gate, publish the ones that pass, and audit your pages. See How it works.

Which AI assistants does it check?

ChatGPT (OpenAI API with web search) and Gemini (with Google Search grounding). Other assistants are on the roadmap. See Prompts.

Why is each prompt asked several times?

AI answers change on almost every run. One check is noise, so each prompt is asked several times per engine (default 5) and you see the share of runs that name or cite you. See Mention rate.

What counts as "visible"?

A prompt is visible in an engine when your site is named or cited in 50% or more of its runs.

Can AgenticSEO guarantee rankings or AI citations?

No. No tool can. AgenticSEO helps you find gaps, publish useful answer-first content, and measure the result.

Keys and cost

Do I need my own AI key?

No. AgenticSEO runs the AI for you, and it is included in your plan. See AI is included.

Who bills me for AI usage?

The AI is included in your plan. See Plans and billing.

How do I see what I spend?

Open AI usage for tokens and searches per job, ticket and day, then apply your provider's prices. See AI usage.

Are my keys safe?

Connection keys are encrypted at rest with AES-256-GCM. Saved keys show only their last 4 characters in Settings.

Do I need paid SEO tools?

No. Search Console, Google Autocomplete, Bing Webmaster Tools and built-in screenshots are free. DataForSEO and paid SERP services are optional. See Keyword sources.

Content and QA

Will it publish without my approval?

Only if you turn on Publish automatically after QA passes. With it off, passed drafts wait in Ready for you. You can also publish as Draft in the CMS. See Publishing.

Can a draft that failed QA be published?

No. The app checks the latest QA report before every publish and refuses drafts that did not pass. See QA gate.

Why did my draft fail QA?

Open the ticket and read the QA report. Most fails come from a product claim that is not in your approved facts, a missing CTA, or the reading grade. See Troubleshooting.

What does "Needs human" mean?

QA failed more times than your rewrite limit (default 3 rewrites), or the ticket is technical work. Fix the cause, then click Rewrite and QA again.

Can the agents make up facts about my product?

They are told not to, and QA fails any business claim that is not in Approved product facts. Product numbers are also checked by code. Keep your facts complete and current.

How do I make the writing sound like us?

Fill in Brand voice and Banned words and phrases in Settings, and add Writer lessons when you see something you would change.

Can someone review a draft without an account?

Yes. Create a review link on the ticket. The reviewer can read the draft and add notes but cannot change it. See Review links.

Setup

Which CMS platforms are supported?

WordPress, Ghost, Webflow, Shopify, a GitHub static site, a custom webhook, or manual publishing. See CMS.

Do I need Search Console?

It is strongly recommended. Research works better with it, and audits and the Search performance page need it. See Search Console.

Can I run more than one site?

Yes. One account can hold many sites, each with its own connections and settings. Your plan sets how many.

How do I stop an agent?

Click Stop on the ticket or on Agent activity. It stops after its current step, and work saved before that is kept.

When do the agents run?

By default, keyword research runs on Monday and the audit on Thursday at 06:00 UTC. The AI prompts check runs right after research. Search Console numbers are pulled daily. Change it in Settings > Weekly schedule.

Tip: Can't find your answer? Check Troubleshooting, or read the agent log on Agent activity.