Training your chatbot: URLs, files, and manual knowledge

A chatbot is only as good as its knowledge. If it doesn't know your product, pricing, or policies — it'll hallucinate answers and frustrate customers.

QuBold gives you three ways to train your bot, each suited to different kinds of content.

1. URLs (crawled automatically)

Paste your website URL and QuBold crawls it: product pages, FAQs, documentation, blog posts. Any publicly accessible page becomes part of the bot's knowledge.

**Best for:** product docs, public FAQs, pricing pages, feature descriptions

**Tip:** Re-crawl URLs whenever your content changes. QuBold shows the last crawl date in the knowledge panel.

2. File uploads

Upload PDFs, Word docs, or plain text files. QuBold extracts and indexes the content automatically.

**Best for:** internal SOPs, product manuals, policy documents, onboarding guides

**Supported formats:** PDF, DOCX, TXT, CSV, XLSX

3. Manual Q&A

Write specific question-and-answer pairs directly. This is the most precise control — you define exactly what the bot says for specific questions.

**Best for:** pricing (which changes), promotions, return policies, escalation scripts

**Example:** - Q: "What's your refund policy?" - A: "We offer a 30-day no-questions-asked refund. Contact support@yourcompany.com."

Combining all three

The strongest bots use all three sources together. URLs for broad product knowledge, files for detailed documentation, manual Q&A for precise answers on sensitive topics.

QuBold merges all sources into a single retrieval index — your bot searches across all of them simultaneously when answering a question.

Testing your bot

After adding sources, use the preview panel to test questions. If the bot gets something wrong, add a manual Q&A entry to override it — manual entries always take priority over crawled content.