Modules · Meeton Chat · Step 3
Knowledge
What the agent reads before it answers. Most teams only need the first three sources: websites, Q&As and files. Everything here is searched per question, so a Q&A nobody asks about costs nothing.
Sources
| Source | How it is used |
|---|---|
| Websites | Crawled pages: the agent answers from them and links to them. The crawl fetches your public pages over ordinary HTTPS like a visitor would, so a bot block or geo block on your site can refuse it. Up to a set number of sites per agent. |
| FAQ (Q&As) | Questions you answered yourself, served in your words, in the visitor's language. When a visitor asks one of these questions, the agent uses your answer instead of writing its own. Prices, dates, names and links are always kept exactly as you wrote them. |
| Files | Documents the agent reads in full. |
| Library documents | Managed on the Library page, not here. The agent can explain and recommend them. |
| More sources (rarely needed) | The database your structured files compile into, which the agent queries for exact values; CRM knowledge articles; and direct knowledge, text placed straight into the prompt, for DynaMeet staff setting a team up. |
Instant answers
When a visitor asks a close match to one of your Q&As, your answer is sent straight away, exactly as you wrote it, translated if their browser language differs. Anything less than a close match still goes to the agent, which reads your answer and replies in its own words. Each Q&A has its own on/off switch, and the step shows how many are on and roughly how many more you can add.
Browse and test
Browse & test knowledge opens the explorer: type a question and see which sources match it and how, the way the agent would. For compiled data, the explorer also lets you write your own read-only query across the objects.
Write Q&As in whichever language you like. The agent replies in the visitor's own browser language and translates your answer on the way out, when auto-translate is on under Advanced.