What changed, and when

The API and MCP now return much more of what you see in the dashboard, including trends over time, results by tag, several projects side by side and the raw data behind every answer. So your own reports, spreadsheets and BI tools can work with the same numbers as the dashboard.

Selected projects now also track the AI answer Bing shows above its results in your market, as a separate assistant. The Answers screen shows how often Bing gave an AI answer at all, and the beta stays out of your "All engines" numbers, so nothing you already track moves.

One list of what to do next to get into AI answers, in four goals: fix how AI reads your site, improve your own pages, get mentioned elsewhere and correct what assistants get wrong about you. Every task points at a specific page, topic or engine. Mark it done, dismiss it, or press Do it and the agent starts the work.

Open an agent draft or load any page from your site by its address, and edit it beside a live score of how ready it is to be quoted by AI and what the assistants really answer on its questions. Fix a section, add an FAQ or rework the whole text with one instruction - every AI change is highlighted until you keep or undo it.

Ask Claude, ChatGPT, Cursor or Gemini CLI about your brand in plain words and get answers built from your real measurements, not from screenshots. You sign in once; anything that would change your data shows a plan first and waits for your confirmation. Included from Growth up and during the trial.

It reads your project's own measurements and drafts what is missing: an article for a question you lose, a rewrite of a page that is not cited, a pitch to a site the assistants trust. Every session ends with a to-do table, and nothing is published without you.

Copy-ready prompts for the questions teams actually ask - a monthly client report, what changed in the answers between two days, which cited sites you could get onto - grouped by task. Paste one into your assistant and it runs on your own data.