Not all chatbots are created equal. Some frustrate. Others win awards.

Some bots solve your problem. Others make you smile while doing it. And then there’s BRUce and Aura; chatbots so well-designed, they were crowned #1 and #2 in Campfire AI’s ranking of Belgium’s best. Behind these digital dynamos? Michiel Baets (Product Owner at Brussels Airport’s BRUce), Toby Moors (Aura’s orchestrator at Loop Earplugs), and David Martens (Ethical AI Professor and Co-Founder of Searching Pi).

In our latest webinar, these three minds met to ask - and answer - the million-euro question: what separates a bot that simply works from one that wins hearts?

If you’re in the business of automation, AI, or just tired of screaming “agent, please” at your screen - keep reading.

Here’s what stuck with us after the webinar:

1. The tone of voice test: your bot isn’t a renegade

Ever had a chatbot say something quirky, only to be contradicted five seconds later by a very serious human agent? That’s brand dissonance - and it’s a killer. Toby made it clear: consistency is key. Your chatbot can have a personality (even multiple ones!), but it should never go rogue. The tone of voice must be aligned across all channels - chatbot, email, human, carrier pigeon. Brand identity doesn’t get a lunch break.

2. Guardrails are KEY; they're more than digital bumpers

Imagine if your chatbot promised a customer a discount… and then your legal team found out you never offered one. (Hi, Air Canada.) David warned us: guardrails aren’t just helpful - they’re essential. These include defining what your bot can and can’t talk about, when to escalate, and how to avoid the AI version of improv comedy. Think: “Sorry, I’m only here for questions about Loop Earplugs.” No off-script drama required.

3. Surprise! Delight doesn’t mean derailment

Toby made a provocative point: a bot that doesn’t surprise occasionally is a missed opportunity. But surprise doesn’t mean saying “YOLO” and ignoring your brand book. It means noticing the compliment before solving the query. If someone says “I love your product,” don’t skip to tracking info - lean into the love. Delight can still live within the script. Just give it a seat at the table.

4. Feedback loops are a chatbot’s gym membership

Want your chatbot to be smarter next week? Then let it learn. Loop’s secret sauce? A continuous cycle of review, human coaching, and model improvement. Their customer service agents aren’t being replaced - they’re being re-skilled as AI whisperers. And yes, they’ve got metrics, dashboards, and sentiment trackers to back it all up.

5. AI that knows its limits wins trust

David dropped a truth bomb: transparency isn’t optional. It’s the law (hello, AI Act). Customers must know they’re chatting with a bot. Why? Because automation bias is real. People tend to trust AI too much - until it makes a mistake. Clear communication keeps expectations (and lawsuits) in check.

6. Intent + GenAI = the hybrid heroes

Both BRUce and Aura started life in the intent-based chatbot era: structured, safe, but often stiff. Now? They’ve evolved into hybrids: part rule-based, part generative. This means better conversations, smarter answers, and fewer “let me transfer you to a human” moments. But the foundation - intent training and FAQ structure - still matters. You can’t build skyscrapers on sand.

7. Voicebots: charming or dystopian?

Imagine whispering sweet nothings to your earplug. That’s where voicebots might be heading. Michiel and Toby dream of bots that notify you when decibels are dangerous - like a guardian angel in your ear. David, meanwhile, waved a caution flag. The deeper the voice connection, the blurrier the human-machine boundary. And no one wants a dystopian 'Her' movie sequel playing out in customer service.

8. Chatbots: where art meets science

This might have been the most profound point of the session. David and Toby said it best: designing a great chatbot is part science (data, metrics, logic) and part art (tone, personality, brand feel). The best bots aren’t just smart - they’re loveable, useful, and aligned with the business. You don’t just launch and leave. You train, test, update, refine… then do it all again.

🎬 Want the full story (and a few chatbot bloopers)?
Leave your info below and you'll receive the recording.

If you’re wondering how to turn your chatbot from frustrating to fan-favorite? That’s exactly what we do at Seven Beaufort. We help ambitious teams design AI experiences that actually make sense—and make money.

Ready to build a bot that breaks the mold? You know where to find us.