Kustomer and the EU AI Act: does your chat widget tell people it's AI?
Kustomer markets its customer-facing AI under the name "Concierge," part of a suite (AI Agents for Customers) built around a shared customer timeline rather than a standalone chat widget. Kustomer leans hard on human-in-the-loop design: Concierge hands off to a human agent with full context, and a human can hand the conversation back to Concierge afterward. That bidirectional handoff is a genuine UX strength, and a genuine Article 50(1) question, since a visitor bouncing between Concierge and a rep needs to know, at each point, which one they're talking to.
The rule itself is short. Article 50(1) of the EU AI Act requires AI systems that interact directly with people to be designed so users know they're dealing with AI, no later than the first interaction, in a clear and distinguishable way. The exception for cases where it's "obvious" is narrow: a natural-language customer-service bot doesn't qualify just because it has a robot icon. A line in your terms of service doesn't satisfy it either. And it's easy to enforce, because a regulator can open your site and start a chat.
Where the disclosure lives in Kustomer
- Your site or app's messaging entry point where Concierge is configured to answer first
- Concierge's presented identity in the conversation (branding, name) as seen by the customer
- Concierge's opening automated response, generated before a human rep is involved
- The two-way handoff points, Concierge to human, and human back to Concierge, where the active participant changes
Your Kustomer disclosure checklist
- Open your support channel as a visitor and start a conversation: does anything identify Concierge as AI before or in its first reply?
- Confirm Concierge's customer-facing branding (not just its internal Kustomer dashboard name) carries an AI descriptor, since "Concierge" alone reads as a human role title
- Add an explicit AI statement to Concierge's opening message rather than relying on the product name
- Because Kustomer supports handing conversations back to Concierge after a human reply, make sure each handback is announced too, not just the initial handoff to a human
- Audit every channel Concierge is deployed to (web, in-app, email-adjacent messaging), since timeline-based platforms often extend one AI configuration across several surfaces at once
- Screenshot the first-interaction state per channel and keep Concierge's configuration export dated
Watch out for
"Concierge" is a name built to sound like white-glove human service, on a platform whose signature feature is passing conversations back and forth between AI and humans. Both are good product decisions and both cut against Article 50(1): the smoother the AI-human handoff feels, the easier it is for a visitor to lose track of which one they're talking to, exactly the clarity the disclosure duty requires.
Common questions
Does Kustomer's timeline model (no separate "chat window") change how disclosure works?
What counts is the interaction itself, whatever the interface is called. Wherever Concierge is the one responding to a customer message, that reply needs to be identifiable as AI, timeline-based messaging doesn't get a pass just because it doesn't look like a classic chat bubble.
We only use Concierge for simple FAQ-style deflection, is that lower risk?
Scope doesn't change the duty, only the stakes if you skip it. If Concierge answers a customer directly, even for simple questions, Article 50(1) treats that as an AI interaction requiring disclosure at first contact.