Duckie's AI support agent now reaches out first. Proactive messaging turns stalled checkouts, failed payments, and silent exits into resolved conversations before customers ever open a ticket.
Support chat has always been reactive: a bubble sits in the corner of a website waiting for someone to click it, and most visitors never do. They hesitate over a pricing tier, stall at checkout, or give up on a setting they cannot find, and then they close the tab without leaving any record that it happened.
We have spent years getting good at the other half of that exchange, so that when someone does ask, Duckie's AI support agent answers the question, looks up the order, resets the password, or checks the plan. But the people who ask have always been the easy ones, and the larger, more valuable group is everyone who got stuck and never said anything at all.
Today we are releasing proactive messaging, which lets Duckie open the conversation rather than wait to be found.
Why it's different
Proactive messages are not new; Intercom and Drift have offered them for years. The trouble is that those messages are scripted, so the bot fires a canned line, the visitor replies with a real question, and the software has nothing to say, having promised a conversation it was not able to hold.
Duckie's proactive chat comes from the same AI customer service agent that already resolves your tickets, carrying the same knowledge base and the same tools, so that when it reaches out about a stalled checkout it can find the failed charge and fix it on the spot. Anyone can start a conversation; the value lies entirely in being able to finish it.
Where proactive customer service helps
The trigger is almost always a signal you already track, whether that is time on a page, scroll depth, a click, an in-app event, or an error, which means you don't need new infrastructure so much as a way for the AI support agent to act on what you are already measuring.
On the revenue side, proactive customer engagement amounts to meeting people at the moment their intent is highest, so that a shopper comparing plans gets help choosing one, a checkout heading toward cart abandonment gets a nudge before the sale is lost, and a trial approaching its limit becomes a conversation about upgrading.
On the support side, it means catching problems before they harden into tickets, so that a failed payment turns into a chance to offer a fix first, a confusing step in the documentation becomes a place to ask for help, and a new signup can be guided through the first real action that so often separates the customers who stay from the ones who quietly leave.
Proactive, not pushy
Outreach handled badly is just spam, and most teams are right to be wary of it, which is why the restraint is built into the product rather than left to good intentions: the agent stays quiet when a visitor already has a conversation underway, defaults to the launcher teaser instead of a popup, and is rate-limited per visitor, on the principle that knowing when not to speak is as much a feature as knowing when to.
The larger point
Taken together, this quietly changes what Duckie is, because an AI support agent that merely handles inbound tickets is a tool for a support team, whereas an AI customer service agent that proactively engages customers across the entire journey and resolves their problems on the spot is something that growth leaders, and whoever owns customer retention, care about too. It is the same product in a considerably larger room.
If you already use Duckie, the simplest way to see what proactive customer support does is to put a single trigger on one busy page and watch what happens.