Docs/train knowledge

Train knowledge

Add URLs, files, and manual Q&A to build the most accurate knowledge base for your chatbot.

Knowledge sources

AskGPT supports multiple sources for training your chatbot. You can combine them to build a comprehensive knowledge base — every source is chunked, embedded, and searched the same way.

→Website URL crawl — auto-discovers all linked pages
→File upload — PDF, TXT, and Markdown documents
→Plain text — paste FAQs, docs, or product info
→Q&A pairs — write question/answer pairs directly

Recrawl your website

When your website content changes, trigger a recrawl from the Train tab. The chatbot will re-index all pages and update its knowledge base. Previous indexed content is replaced.

Manual Q&A pairs

Use Q&A pairs to add answers for questions your website doesn't explicitly answer — pricing, custom processes, policies, or edge cases your visitors ask about.

💡Manual Q&A pairs have the highest priority and will always appear in search results over crawled content.

Credit usage

Crawling and indexing uses credits: 10 credits per crawl job, 1 credit per content chunk indexed. Manual Q&A pairs do not use additional credits once saved.

Create chatbot Test and publish

Need more help?

Contact support and we'll get back to you within 24 hours.