See it in action
Add your content, keep it current, then see exactly which answers are missing.
Connect your site, files, and docs, then see exactly which answers are still missing.
Grounded in your content
What the Knowledge Base does
Your AI answers only from sources you add, stays current on its own, and shows you exactly what it still can’t answer.
Reads your whole site
JavaScript-heavy pages, single-page apps, and protected storefronts get read in full, so every product and page makes it into the answers, not just the easy ones.
Bring everything else in
Upload PDFs, Word, and text files, or write Q&A answers by hand for the knowledge that lives nowhere else on your site.
Notion Synced
Google Docs Synced
Zendesk SyncedNo outdated answers, ever
Sources re-sync on a schedule you set, and Notion, Google Docs, and Zendesk update live, so retired products and old prices never reach the chat.
The right answer, not just any answer
Every question searches your full knowledge base and surfaces the most relevant match, so visitors get the precise answer instead of a close guess.
Separate sources, no crossed wires
Mark any source as Copilot-only to keep it out of your visitor-facing AI Agent. Customers never see what’s not for them, while your team still can.
Shows you what’s missing
Spot the questions visitors ask that your AI can’t answer yet, so you know exactly what to add next.
How it works
From setup to a resolved conversation.
Add your sources
Enter your website URL, upload files, write Q&A entries, or connect Notion, Google Docs, or Zendesk.
We read and organize it
Your pages, files, and synced docs are read, organized, and made searchable, so the AI can pull the right source for every answer.
Answers stay current
Re-crawls and syncs keep the knowledge base fresh; the AI cites its sources on every answer.
Knowledge Base FAQ
How BubblaV knowledge base works.
What can I train the AI on?
What can I train the AI on?
Your website (crawled with full JavaScript rendering), uploaded files (PDF, Word, TXT, Markdown up to 10MB), manual Q&A entries, and live-synced content from Notion, Google Docs, and Zendesk help-center articles and tickets.
How does BubblaV handle JavaScript-heavy sites?
How does BubblaV handle JavaScript-heavy sites?
The crawler renders single-page and JavaScript-driven sites with full browser-level rendering, works through bot protections using realistic browser headers, and extracts ecommerce product catalogs, so shops that render client-side still get crawled.
How are answers kept accurate?
How are answers kept accurate?
Retrieval uses hybrid vector + full-text search with reciprocal rank fusion and reranking, and the AI cites its sources. If a question isn't covered, it can escalate instead of guessing, and content-gap reports show you exactly what to add.