Get your Help Center found in search engines and LLMs

July 6, 2026

Summary Your Help Center is an auto-generated collection of articles, written entirely by AI and kept up to date from your self-updating Living Knowledge. This release turns it into a discovery channel. You can now publish it as fast, standalone pages on your own subdomain, so search engines and LLMs can crawl, index, and cite your articles, and the people (or AI assistants) looking for those answers reach them directly, often before a ticket is ever opened. Clean, readable links and stable deep links make every page easy to share, index, and reference.

Alongside the Help Center, the agent gets better at handling unclear questions by asking a couple of targeted follow-ups before it answers. A range of smaller improvements and fixes then tidy up content titles, procedure data, AI maturity tracking, and live chat.

New Features

• Publish your Help Center as pages search engines and LLMs can read. You can now publish your Help Center as fast, standalone pages on your own subdomain, so search engines and AI assistants can index and cite your articles, and people find your answers directly, often before they open a ticket.
• Sharper answers to unclear questions. When a question could mean several things, the agent now asks a couple of quick follow-up questions and then gives the exact answer, instead of guessing from the wrong article.

Improvements

• Deep links in the Help Center. Each category and article has its own stable address that updates as visitors navigate and survives a refresh or the back button, so every page can be linked, shared, and indexed on its own.
• Readable, search-friendly article links. Help Center articles now use clear, word-based URLs instead of a string of characters, which read better when shared and help search engines and LLMs understand each page.
• Faster Help Center answers. Answers drawn from a Help Center article are now cached and served almost instantly, refreshing within a day or right away when you republish.
• Clearer Help Center admin. Labels and layout across the Help Center admin are tidied up, including renaming “Questions” to “Articles” to match how the Help Center is organized.
• Unpublish whenever you need to. A remove button next to publish takes your live Help Center offline at any time, without touching your configuration.
• Open the agent on a specific article. A new API opens the AI component directly on a chosen Help Center article, so you can send people straight to the right answer.
• Better titles for library content. Content library articles now show a properly generated title, including special characters, instead of one pulled from the file path.
• Jump straight to a source. Opening a content library source from an interaction now takes you directly to that content node’s page, instead of a popup.
• Retrain after connecting your content library. You can now retrain your quality questions once the content library is enabled, so their answers reflect your latest content.
• Clearer procedure data. A new Txt.setProcedureData() API lets you pass data into a procedure, and what used to be called chat metadata and variables is now consistently called procedure data.
• Focus your AI maturity by stage. You now activate a lifecycle stage before scoring it, so your maturity view reflects the stages you are actually working on.

Bug Fixes

• Live chat stays in sync. Opening a second chat in live chat mode no longer lets the two fall out of sync.
• Procedure pages reload correctly. Refreshing while on a procedure’s detail page now loads the procedure properly instead of an incomplete view.
• Maturity tasks match your focus. Tasks are no longer created for a lifecycle stage you have not chosen to focus on.
• No needless title rewrites. A Help Center article’s title is no longer rewritten when it is already in the same language.
• Cleaner replies to greetings. The read more link now appears only after the agent has actually answered a question, not after a simple greeting or “how can I help” reply.
• Longer action names handled gracefully. Resolved an error that could occur when an action’s name was too long; names now stay within the supported length automatically.