Your first Autopilot chat

In shortAutopilot is a chat agent (labelled Beta) that runs AI Rank Lab tools for you and hands back reports, spreadsheets, documents and dashboards. Open Content → Write → Autopilot agent (it opens in a new tab), pick a website, describe the task and press send. Before it uses any paid tool, it shows an estimated credit spend for you to confirm. You pay for model use plus each tool's own price.

What Autopilot is for

Most AI Rank Lab tools do one job each. Autopilot chains them. Ask "audit my pricing page, research keywords for it and give me a spreadsheet of fixes" and it plans the steps, calls the audit and keyword tools, reads the results and writes the answer, with files you can download.

Use it when a task needs several tools, or when you would rather ask than click. For a single, familiar job, the tool's own page is usually quicker and shows you more detail.

Before you start

• At least one website added. Autopilot works on the site you pick; if you have none, the website picker sends you to Websites to add one.
• Enough credits for the task. If the estimate is more than your balance, Autopilot stops before it runs any paid tool and suggests a smaller version.
• Optional: Google Analytics 4 and Search Console connected, if you want answers from your real traffic. Autopilot asks you to connect them when a task needs them.

Step by step

1. Open AutopilotIn the dashboard sidebar, click Content → Write → Autopilot agent. It opens in a new browser tab, with its own sidebar of chats and tools. The BETA label next to the name is a fair warning: it is newer than the rest of the product.
2. Pick the websiteUse Select website above the message box to choose which of your sites this chat is about. The agent uses it as context for every step.
3. Describe the taskType in the box (How can I help you today?), or click a starter card such as Full site audit, Keyword CSV export or AEO opportunity report. The options menu next to the box has Start from… shortcuts and an Effort Mode: Auto for balanced speed and depth, or Deep dive for a thorough multi-tool analysis that uses more credits.
4. Confirm the spendAutopilot reads your request and plans the tools it will use. If the plan includes any paid tool, a Confirm credit spend card shows the estimated total and a line per tool. Click Proceed to run it, or Cancel and ask for something smaller.
5. Watch it workWhile it runs you see a status line for each tool call as it happens, then the written answer. Use the stop button to cancel. Closing the tab doesn't stop the run; reopen the chat later to see the result.
6. Collect the resultsDashboards and documents open in the panel on the right. Spreadsheets, Word files, PDFs and images appear as file cards you can preview or download. All files are also kept in Media. Use Share this chat to create a read-only link for a colleague.

How you pay

Each turn has two parts. Model use is charged by the number of tokens the agent reads and writes, at the end of the turn, with a small minimum. Tools are charged at a fixed price per call, before the call runs, and refunded if the call fails. Reading your own data, fetching a public page, making a dashboard and asking you a question are free tools. Next to the chat title, a small counter shows the credits and model tokens that chat has used so far.