{"Status":true,"Message":"","Response":{"post":{"postuid":"5fdbe1a5-6ff3-4a4e-9766-851886def654","tenantuid":"d8b744fc-2e70-4089-bb80-dd1d08f6c7b2","projectuid":"fc6490ac-7527-4f49-b06e-46f701280e85","title":"AI Answers on Your Help Site","slug":"article/ai-answers-on-your-help-site","html":"\u003Cp\u003EReaders can ask your help site a question in their own words. HelpGuides finds the most relevant published articles, writes an answer from them, and shows the articles it used. There\u0027s nothing to set up: every help site has AI answers.\u003C/p\u003E\u003Ch2 id=\u0022where_readers_ask_questions\u0022\u003EWhere readers ask questions\u003C/h2\u003E\u003Ctable border=\u00221\u0022 style=\u0022border-collapse: collapse; width: 100%;\u0022\u003E\u003Cthead\u003E\u003Ctr\u003E\u003Cth\u003EWhere\u003C/th\u003E\u003Cth\u003EHow it works\u003C/th\u003E\u003C/tr\u003E\u003C/thead\u003E\u003Ctr\u003E\u003Ctd\u003ESearch popup\u003C/td\u003E\u003Ctd\u003EClick \u003Cb\u003EType ctrl-k to search\u003C/b\u003E in the sidebar, or press Ctrl\u002BK (\u2318K on a Mac). The popup is for questions: every question typed into \u003Cb\u003EAsk a question ...\u003C/b\u003E gets an answer. Before anything is asked, it shows up to five \u003Cb\u003ERecently updated\u003C/b\u003E articles.\u003C/td\u003E\u003C/tr\u003E\u003Ctr\u003E\u003Ctd\u003E\u003Cb\u003EAsk AI about these docs...\u003C/b\u003E\u003C/td\u003E\u003Ctd\u003EThe box at the top of your docs home page and each category page. A question asked here opens the search popup with the answer.\u003C/td\u003E\u003C/tr\u003E\u003Ctr\u003E\u003Ctd\u003E\u003Cb\u003EType to search\u003C/b\u003E\u003C/td\u003E\u003Ctd\u003EThe search box above an article. It lists matching articles as you type, and answers the query when it looks like a question.\u003C/td\u003E\u003C/tr\u003E\u003C/table\u003E\u003Cfigure\u003E\u003Cimg class=\u0022lazy-load\u0022 data-src=\u0022https://graffiti-auf7e6dwhxhcbwek.z03.azurefd.net/d8b744fc-2e70-4089-bb80-dd1d08f6c7b2/fc6490ac-7527-4f49-b06e-46f701280e85/images/0df7241700d94c378688851bb8b45516.png?v=843942633\u0022 data-width=\u00221300\u0022 data-height=\u0022760\u0022 style=\u0022max-width: 100%;\u0022 alt=\u0022Search popup asking for a question with recently updated articles below\u0022/\u003E\u003Cfigcaption\u003EThe search popup\u003C/figcaption\u003E\u003C/figure\u003E\u003Cfigure\u003E\u003Cimg class=\u0022lazy-load\u0022 data-src=\u0022https://graffiti-auf7e6dwhxhcbwek.z03.azurefd.net/d8b744fc-2e70-4089-bb80-dd1d08f6c7b2/fc6490ac-7527-4f49-b06e-46f701280e85/images/3cc00b67639f47f0a4bc8a45c934f9da.png?v=825944894\u0022 data-width=\u00221300\u0022 data-height=\u0022900\u0022 style=\u0022max-width: 100%;\u0022 alt=\u0022Docs home page with the Ask AI about these docs box\u0022/\u003E\u003Cfigcaption\u003EAsk AI on the docs home page\u003C/figcaption\u003E\u003C/figure\u003E\u003Cp\u003EIn the \u003Cb\u003EType to search\u003C/b\u003E box, a query gets an answer when it ends with a question mark or starts with one of these words followed by a space: who, what, when, where, why, how, is, are, can, do, does, should or will. Anything else returns a list of matching articles.\u003C/p\u003E\u003Ch2 id=\u0022what_an_answer_looks_like\u0022\u003EWhat an answer looks like\u003C/h2\u003E\u003Col\u003E\u003Cli\u003EThe reader\u0027s question appears, and the answer streams in beneath it, formatted with lists, bold text and links.\u003C/li\u003E\u003Cli\u003EWhen the answer is complete, the articles it was written from are listed as sources, up to three.\u003C/li\u003E\u003Cli\u003EUp to three suggested follow-up questions appear. Clicking one asks it.\u003C/li\u003E\u003Cli\u003EA \u003Cb\u003EDidn\u0027t find what you were looking for? Let us know ...\u003C/b\u003E link lets readers tell you the answer didn\u0027t help.\u003C/li\u003E\u003C/ol\u003E\u003Ch2 id=\u0022followup_questions\u0022\u003EFollow-up questions\u003C/h2\u003E\u003Cp\u003EAfter an answer, the popup\u0027s box changes to \u003Cb\u003EAsk a follow-up ...\u003C/b\u003E. Follow-ups are understood in context, so a reader can ask \u0022does that work on mobile?\u0022 without repeating the topic. The last three questions and answers are sent along with each follow-up. Nothing is stored between visits: closing the popup starts a new conversation.\u003C/p\u003E\u003Ch2 id=\u0022what_the_ai_uses\u0022\u003EWhat the AI uses\u003C/h2\u003E\u003Cul\u003E\u003Cli\u003E\u003Cb\u003EPublished articles only.\u003C/b\u003E Drafts and scheduled articles are never used.\u003C/li\u003E\u003Cli\u003E\u003Cb\u003EThe best matches.\u003C/b\u003E Articles are indexed in sections, split at headings, and the answer is written from the three best-matching articles.\u003C/li\u003E\u003Cli\u003E\u003Cb\u003ENo outside knowledge.\u003C/b\u003E The AI is told to answer only from those articles and not to make anything up. If they don\u0027t cover the question, it says it doesn\u0027t have enough information to answer.\u003C/li\u003E\u003C/ul\u003E\u003Cp\u003EThe best way to improve an answer is to improve the article behind it: clear headings, one topic per section, and the words your readers use.\u003C/p\u003E\u003Ch2 id=\u0022keyword_search\u0022\u003EKeyword search\u003C/h2\u003E\u003Cp\u003ESearch matches by meaning by default, so \u0022change my password\u0022 also finds \u0022reset your login\u0022. To match exact words instead, start the query with \u003Ccode\u003Ekeyword:\u003C/code\u003E in the \u003Cb\u003EType to search\u003C/b\u003E box:\u003C/p\u003E\u003Cdiv class=\u0022code_wrapper\u0022\u003E\u003Cdiv class=\u0022code\u0022 data-language=\u0022text\u0022\u003Ekeyword:webhook\u003C/div\u003E\u003C/div\u003E\u003Ch2 id=\u0022see_what_readers_search_for\u0022\u003ESee what readers search for\u003C/h2\u003E\u003Cp\u003EThe \u003Ca href=\u0022/article/dashboard\u0022\u003EDashboard\u003C/a\u003E shows \u003Cb\u003ERecent Searches\u003C/b\u003E, the searches where a reader clicked through to an article, and \u003Cb\u003ESearch None Helpful\u003C/b\u003E, the searches where a reader clicked the \u003Ci\u003EDidn\u0027t find what you were looking for?\u003C/i\u003E link. Use them to find missing or unclear articles.\u003C/p\u003E\u003Ch2 id=\u0022the_same_answers_elsewhere\u0022\u003EThe same answers elsewhere\u003C/h2\u003E\u003Cul\u003E\u003Cli\u003EIn Slack, with \u003Ccode\u003E@HelpGuides\u003C/code\u003E. See \u003Ca href=\u0022/article/connecting-slack\u0022\u003EConnecting Slack\u003C/a\u003E.\u003C/li\u003E\u003Cli\u003EOn any website, with the \u003Ca href=\u0022/article/helpguides-chat-overview\u0022\u003EChat messenger\u003C/a\u003E.\u003C/li\u003E\u003Cli\u003EIn AI assistants, through \u003Ca href=\u0022/article/model-context-protocol\u0022\u003EModel Context Protocol\u003C/a\u003E.\u003C/li\u003E\u003Cli\u003EThrough the public search API described by \u003Ca href=\u0022/article/exposing-your-help-api-with-openapijson\u0022\u003Eopenapi.json\u003C/a\u003E.\u003C/li\u003E\u003C/ul\u003E","publish_status":0,"post_type":"Article","authoruid":"3dde8c16-763a-4a2b-ae0b-1d8c50c62e3d","author":{"authoruid":"3dde8c16-763a-4a2b-ae0b-1d8c50c62e3d"},"featured_image_updating":false,"meta_description":"HelpGuides lets readers ask questions naturally and AI provides answers from your published articles, enhancing help sites with instant, sourced responses.","display_toc":true,"has_workingcopy":false,"allow_indexing":true,"total_views":1,"date_published":"2026-10-04T21:50:03.49","date_updated":"2026-10-04T21:50:04.16","date_created":"2026-09-28T16:21:14.19"}}}