The questions we get asked

No. The console gives you one line to paste into your site, and step-by-step instructions for WordPress, Wix, Squarespace and Shopify. If you can edit your own pages you can have Evrella installed in a couple of minutes.

No. She only answers from your website and the documents you give her, and every answer links to the page it came from. If the answer isn’t there she says so instead of guessing. That’s usually a sign you have a page missing.

Whatever language the visitor writes in. Someone asking in Polish gets a reply in Polish, from the same English pages you already have. Your enquiry emails and your console stay in your language.

By email, the moment someone leaves their details, with their name, their contact details and the conversation behind it. You can have it by text as well if you give us a UK mobile number. The same enquiry also waits in your console with the whole exchange, so nothing hangs on one message getting through.

Nothing for 30 days, and you don’t need a card to start. Pick a plan, try it, and add a card before day 30 to keep Evrella answering: your first charge is on day 30. If you don’t, she pauses and nothing is deleted. Plans are £49 or £99 a month, no setup fee, no VAT.

Add 100 for £10 from the console, or move up a plan. Evrella stops when you hit the limit rather than quietly running up a bill, which is why you can leave her switched on without worrying.

Booking, yes: it is in every plan. WhatsApp comes with Growth, and we set it up with you. Answering the phone is a custom set-up on the Custom plan: we connect her to your number with you, after a conversation instead of a checkout.

Take the line of code out of your site and she stops straight away. There’s nothing to uninstall. Your plan is cancelled from the console’s Plan & billing page and won’t renew.

On AWS in London (eu-west-2): the application, its databases and its backups. We won’t tell you it never leaves the UK, because that wouldn’t be true. Incoming email is handled in Ireland, and the model that writes each answer is called through an EU inference profile. All of this, along with our sub-processors and how long we keep things, is set out on the trust page.