It asks qualifying follow-up questions before it names anything. Size, use, budget, context, whatever your own content actually distinguishes on. The recommendation comes from your catalogue as you have written it, not from the open internet.
It answers buying questions from your own content, narrows the choice, and moves the visitor to the next step you already have.
Buying questions arrive on a product page, a pricing page or a comparison page. The visitor wants to know which option suits them, what separates two similar items, and whether it fits their situation. Blengi answers there, inside the page.
It is trained only on your own content: your website and sitemap, PDFs, Word files, Google Docs, Notion pages, pasted text and SQL databases. Sources are re-crawled automatically, so the answers follow your content. You can also re-index a source on demand when a change has to land now. When your content does not cover something, the assistant says so.
It asks qualifying follow-up questions before it names anything. Size, use, budget, context, whatever your own content actually distinguishes on. The recommendation comes from your catalogue as you have written it, not from the open internet.
Your plan and feature detail lives in pages the assistant has read, so it can set two options side by side and explain the difference. It answers in the visitor's language automatically. For wording that has to be exact, pin a curated answer and that is what gets repeated.
Run it as a floating bubble, a side panel, or an inline block placed inside the page itself. Suggested starter questions give the visitor an opening that matches the page they are on, so the conversation begins on the thing in front of them.
Being clear about the boundary is part of the design.
stappsokken.nl runs Blengi on its storefront. Installing is one script tag on any platform. There is a plugin for WordPress and WooCommerce, and Google Tag Manager also works.
See how it works