Add your website
Create a website entry in the dashboard and provide the URL you want the chatbot to learn from.
Retrieval-grounded answers
Use retrieval-augmented generation to test conversational answers based on pages from your own website.
A general-purpose language model may not know your current pricing, policies, or documentation. Chatbot Engine first retrieves relevant passages from the content indexed for your website and supplies that context when generating an answer.
This does not guarantee that every answer will be correct. The dashboard gives you a place to inspect the knowledge and test representative questions before relying on the output.
How it works
A straightforward workflow that keeps you in control of the content your chatbot can use.
Create a website entry in the dashboard and provide the URL you want the chatbot to learn from.
Crawl up to 10 pages, then review the indexed content in your knowledge library.
Ask real questions in the dashboard and check that the answers match your website content.
Product capabilities
Relevant indexed content is selected as context for each question.
Use public website pages as the source material for answers.
See the knowledge associated with a website in the dashboard.
Ask natural questions and review the generated responses.
Re-crawl a website when the source content changes.
Keep each chatbot tied to the content indexed for its website.
No. Chatbot Engine is a separate product that uses a conversational language model with retrieval from your indexed website content.
It is a workflow that retrieves relevant source content before asking a language model to generate an answer.
No. Retrieval improves context, but generated answers should still be tested and reviewed for your use case.
Yes. The dashboard includes a knowledge library for the selected website.
The current crawler processes up to ten pages per crawl.
Add a website, review the indexed content, and ask representative questions.