> ## Documentation Index > Fetch the complete documentation index at: https://docs.weelkit.com/llms.txt > Use this file to discover all available pages before exploring further. # Prompts > Add the questions Weelkit asks AI assistants, choose the assistants, review suggestions, run checks, and add answers you saw yourself. A prompt is a question your audience asks AI assistants, such as "What are the best tools for editorial teams?". Weelkit asks each monitored prompt to the assistants you choose, on a schedule, and checks every answer for **Visible**, **Mentioned** and **Cited**. See [AEO & GEO](https://docs.weelkit.com/metrics/aeo-and-geo.md#visible-mentioned-and-cited). Open **Metrics → AEO & GEO → Prompts**. Everyone who can see AEO & GEO can read prompts and their answers. Adding, changing, running and stopping prompts is for Managing Editors and Admins. ## The prompts list Switch between **Monitoring**, **Suggestions** and **Stopped**, and search by prompt. Each view shows its count. In **Monitoring**, each prompt shows: * **Assistants:** the assistants it's asked to. Point at an icon to see the assistant and the method, for example "ChatGPT · API". * **Latest check:** three ticks for **Visible**, **Mentioned** and **Cited**, and how many answers were checked. * **Best position:** your best rank in the prompt's search results. * **Next check:** when it runs next, or **Checking**, **Due now**, or **Overdue**. A check is overdue when it hasn't started within 30 minutes of being due; point at **Overdue** to see why. The strip above the list shows the three rates, **Prompts monitored**, **Suggestions waiting** and **Checks overdue**. When your prompts leave a gap, such as no prompt for one stage of the buyer's journey, the list recommends prompts to add. Monitoring runs either way. ## Add a prompt 1. Choose **Add prompt**. 2. Under **Prompt**, type the question the way a reader would ask it, up to 512 characters. 3. Under **Assistants**, choose who to ask. They're grouped by method: * **API:** the assistant's own API: ChatGPT, Claude, Gemini, Grok, Perplexity, DeepSeek and Kimi. * **Apps:** the assistant's public app, as a signed-out visitor sees it: ChatGPT, Gemini, Google AI Mode, Google AI Overview, Brave AI Answer, Brave AI Mode, DuckDuckGo Search Assist and Duck.ai. Use **Select all** to choose a whole group. The assistants that are ready for your workspace start ticked. 4. Turn on **Priority prompt** to check it on the priority schedule. 5. Choose **Preview wording**. Weelkit writes the wording it will use for each of your languages: what it asks the assistants (**Asked to assistants**) and the searches it runs to check rankings (**Searched as**). Open **Wording preview** to read it. To get new wording, choose **Rewrite wording**. 6. Choose **Add prompt**. A new prompt is checked straight away. Later checks follow the schedule in [Organization settings → AI visibility](https://docs.weelkit.com/settings/organization.md#ai-visibility): one schedule for regular prompts and one for priority prompts. ### Skipped assistants When an assistant can't be asked, its check shows **Skipped**: in the **Assistants** column (greyed out), in **Latest check**, and on the answers. Point at it to see why, for example **Not set up for your workspace**, **App checks off** or **Assistant not available in your market**. A skipped check is never counted in any rate. ### Add a prompt for an article Open a published article in [AEO & GEO → Articles](https://docs.weelkit.com/metrics/aeo-and-geo.md#articles). Choose **Add prompt** on the right of **Monitored prompts**, then complete the same dialog above. You need access to manage prompts. The new prompt is linked to that article. An answer counts as **Mentioned** when it names the article by its exact title, and as **Cited** only when it links that article. ## Work with a prompt Choose **View prompt** from a prompt's **⋯** menu to open its page. It shows the prompt's three rates and how many answers were checked, each assistant's results, and the **Latest check**: its three results, **Your best position in search**, and the competitors that rank. Choose **View ranked results** for the full search results. * **Run now** checks the prompt again straight away. * **Wording sent to assistants** shows the current wording. Choose **Rewrite** to replace it; future checks use the new wording. * **Answers** lists every answer to the prompt. Choose **View answer** to read one in full. * **Edit prompt** (in the **⋯** menu) changes the prompt, its assistants and its priority. ### Stop and start monitoring To stop checking a prompt, choose **Stop monitoring** from its **⋯** menu. It moves to **Stopped**, and its answers stay. To monitor it again, open **Stopped** and choose **Start monitoring**. ## Suggestions Weelkit suggests prompts worth monitoring from your content and your search data, such as the Search Console queries that bring readers to your site. It suggests them automatically on the schedule in [AI visibility settings](https://docs.weelkit.com/settings/organization.md#ai-visibility), or when you choose **Suggest prompts**. A run can find no new prompt worth tracking, and says so. Open **Suggestions** to review them. **Why it was suggested** says why each one is worth tracking. * **Review & monitor** opens the suggestion with its evidence under **Why Weelkit suggested this**: for example the page of yours it matches, the Search Console query and its impressions, and your rank in a live search. Choose the assistants, then **Save & monitor**. * **Dismiss** asks for an optional reason. The suggestion moves to **Stopped** as **Dismissed suggestion**; choose **Start monitoring** there if you change your mind. To get suggestions for one article, open the article in [AEO & GEO → Articles](https://docs.weelkit.com/metrics/aeo-and-geo.md#articles) and choose **Suggest prompts**. A prompt you monitor from an article's suggestions is linked to that article, just like [one you add there](#add-a-prompt-for-an-article). ## Add an answer you saw yourself When you see an AI answer yourself, you can add it to a prompt. 1. Open the prompt and choose **Add manual answer**. 2. Under **Answer**, paste the answer exactly as you saw it. 3. Under **Links it showed**, add the links the answer showed, one URL per line. 4. Choose **Add answer**. Weelkit works out **Mentioned** and **Cited** from the pasted answer and its links, the same way as for any other answer. Manual answers show the method **Manual**, and they count in the rates. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.