Straight answers3 min read

AI chat with human handoff and full conversation context

Short answer

Supportifier's live chat starts with an AI agent that answers from your published, reviewed articles with links to the source, and shows your own fallback message when it does not know. The visitor can choose to talk to a person at any time; the agent then sees the whole conversation and, for an identified customer, the order and delivery data the AI used. A chat nobody picks up becomes an ordinary ticket, and the transcript is saved on it.

The short answer: In Supportifier's website chat the visitor starts with an AI agent that answers only from your reviewed articles, and can ask for a person at any time. The agent who takes over sees the whole conversation and the customer data the AI used, so the customer never repeats themselves. If nobody picks up, the chat becomes a ticket in the inbox with the transcript attached.

How a chat runs

  1. The visitor writes in the chat widget on your site. The AI agent searches your published articles, reads the conversation so far and answers, with links to the articles it used.
  2. If the visitor is identified and a connection exists, the AI also fetches order status, delivery, payment and, for a signed-in customer, loyalty membership, points and vouchers from Voyado, and answers from that.
  3. When the AI does not have the material it does not guess: it shows your own fallback message and offers a person.
  4. The visitor can choose "Talk to a human" at any point, not only when the AI fails.
  5. An agent takes the chat with the full conversation in front of them. If no agent picks it up within a few minutes, the chat becomes an ordinary ticket, and the customer gets the reply by email.
  6. The transcript is saved as a ticket either way, so the history is in the inbox and in the customer portal.

What the agent sees at handover

The whole conversation, the articles the AI used, the customer's earlier tickets and the same customer data cards the AI read: orders, deliveries, returns, payments and invoices from the connected systems. The agent also gets a reply suggestion built from the same sources, fact-checked and labelled Safe, Review or Uncertain, and decides what to send. Customer data is shown to the agent and used by the AI, but never sent out raw and never exposed in the customer portal.

What the AI can do in the chat

For an identified customer, and only where you have switched it on, the AI can cancel an unshipped order, open a return or issue a discount code directly in the chat, within the limits you set, such as the cancellation window and the size of the discount. Each action is off by default and everything is logged.

Outside opening hours and on other channels

The AI chat answers around the clock from the same articles. Outside opening hours, chats that need a person become tickets for the next working day. The same handover logic applies in WhatsApp, Messenger and Instagram Direct, where conversations land in the same inbox, and in the AI contact form, which tries to answer before the question becomes a ticket.

Languages and setup

The AI answers in the visitor's language; help centre, chat and form are available in Swedish, English, Norwegian, Danish, Finnish and German. The widget is embedded on your website with one line of code, there is a Shopify app with theme blocks and a WordPress plugin, and the public help centre can run on your own domain.

What it is not

It is not a scripted bot with decision trees, and it is not a general-purpose chatbot: the AI only uses your reviewed articles and the customer's data. It is also not a replacement for the team. The design is that the AI takes the questions your articles already answer, and people take the rest with the context ready.

Read more on the AI chat page, the help centre page and AI agents that answer from your reviewed knowledge base.

This page is also available in Swedish: AI-chatt med överlämning till människa och hela sammanhanget

Next step

Start from your everyday work.

Tell us which question takes time today. We go through what a first step could look like.

A clear first step beats a big promise.

See how the knowledge work, the review and the first channel fit together.

About the knowledge analysis
From the same question
to a better answer.
Book a walkthrough