Docs/Train

Keeping knowledge up to date

Auto-refresh, refreshing on demand, and managing or retrying individual pages.

Your website changes: prices go up, services are added, opening hours shift. Liyobot can re-read your site on a schedule so your chatbot keeps up, and lets you manage individual pages when you need more control.

In the dashboardChatbots›Your chatbot›Train›Knowledge

Auto-refresh#

Every website source has an AUTO-REFRESH setting in the sources table. Choose how often Liyobot should check your site:

OptionWhen to use it
OffYour site rarely changes, or you prefer to refresh by hand.
DailyPrices, stock or news change often.
WeeklyYou update your site a few times a month.
MonthlyThe default for new websites. Good for sites that change occasionally.

Liyobot regularly checks which sources are due and refreshes them in the background. You don’t need to do anything.

How a refresh works#

A refresh is smarter than starting again from scratch:

  1. It looks for new pages

    Pages you’ve added to your site since the last crawl are found and added, within your page allowance.

  2. It only re-reads what changed

    Pages whose content hasn’t changed are left as they are. Only changed pages are read again, so refreshes are quick.

  3. The bot uses the latest version

    Once a changed page is re-read, answers draw on the new content.

Refresh now

Just updated something important? Click Refresh now (⟳) on the website’s row to refresh immediately, rather than waiting for the schedule. To refresh several websites at once, tick their rows and use Refresh now in the bulk actions bar. A refresh can’t start while a crawl or refresh of that source is already running.

View and manage pages#

Click View pages (▤) on a website’s row to see every page in that source. Each page has a status dot and one of these labels:

LabelMeaning
N chunks · changed n agoIndexed and in use. Shows how many passages it holds and when its content last changed.
re-indexing…Being read again right now.
failed · auto-retry in nCouldn’t be read. Liyobot will try again automatically after the time shown.

Each page has two actions. Re-index reads that one page again, which is handy right after you edit it. Remove takes the page out of your bot’s knowledge after you confirm “Remove this page?”. Removing pages you don’t need also frees up your page allowance.

Failed pages and retries#

If some pages fail, a banner at the top of the panel says “N pages failed · will retry automatically”. Liyobot retries them in the background, waiting longer between attempts. Pages that don’t exist (404) or are private aren’t retried.

If you’ve fixed the problem, for example by bringing a page back online, click Retry N failed now to try straight away.

Tip
If a page keeps failing, open it in a private browser window. If it asks you to log in or shows an error, Liyobot can’t read it either. Remove it, or add its content as a text snippet.

Re-index and re-embed#

Files and text don’t have auto-refresh, because they only change when you change them. For any source, Re-index (on files and text) or Re-embed (in the ⋯ menu) processes its content again. Use these if answers drawn from a particular source seem out of date. You can re-embed several sources at once with the bulk actions bar.

Note
Re-indexing a file reprocesses the same file. If the document itself has changed, upload the new version and delete the old one.

Still stuck? Email [email protected] and a person will help.