Auto-Retrain

Automatically re-crawl URL-based knowledge sources to keep your chatbot up to date

Auto-Retrain

Keep your chatbot's knowledge current by automatically re-crawling URL sources on a schedule.

How It Works

A scheduled background task checks for URL-type sources with auto-retrain enabled (the schedule is attached in the background-job dashboard, typically hourly)
Checks if enough time has passed since the last retrain (default: 24 hours; a source that has never been retrained runs immediately)
Re-crawls the URL
Compares word count — if content changed by more than 10 words, re-indexes the source
If unchanged, skips re-indexing (saves processing)

Sources with auto-retrain enabled show a refresh icon in the Knowledge sources list.

Configuration

Interval: 24 hours (configurable per source via retrainInterval, in hours)
Content change detection: Compares word counts before and after crawl
Skips unchanged content: Only re-indexes when content actually changed
Applies to url, website, and webpage source types

Last updated on

Features

Explore all RKT platform features — skills, handoff, webhooks, automations, and more

Automation Rules

Create no-code "when X happens, do Y" rules for your chatbots

On this page