Onboarding now asks for one thing: your storefront address

Getting started used to mean two screens and half a dozen fields: an organization name, a site name, a platform picked from a list of thirteen, a URL. Now there's one field (your storefront address), and we work the rest out from it.

Submitting it creates your organization, your site, and a working assistant straight away. From your storefront's own pages we take the site name, the language your shoppers read, and the name shoppers will see the assistant under. If we can confirm your platform, a catalog connection is set up too and linked to the assistant from its very first version, so it can search your catalog and read the cart from the first conversation. We only do that when the platform is genuinely confirmed, never on a hunch: no connector is better than one pointing at something we never inspected.

Then we read your store (a web search on your brand plus a fresh read of your own pages) and write a description of what you sell and how your brand sounds. That runs in the background, so you can close the tab. We email you when it's done, in your dashboard language, with a link back to where you left off.

Everything we generated is then shown to you on one screen, and every line is editable. Edits save as you make them. When you're happy, you land in the playground talking to your own assistant instead of staring at an empty dashboard.

If the analysis comes back with nothing usable (a store behind heavy bot protection, say), you still have a working assistant. We just ask for a couple of sentences about what you sell, and you can skip that too.

Nothing reaches a shopper through any of this. The assistant only starts answering real customers once you install it on your storefront or turn on its public link.

Adding a second storefront later uses the same flow, and your existing organization is left alone.