When VCCP and O2 created Daisy, an AI-powered grandmother designed to keep phone scammers occupied, they weren’t just building another AI tool. They were exploring what happens when artificial intelligence is given a distinct personality, purpose, and sense of humour.

We spoke with Ben Hopkins and Morten Legarth about the inspiration behind Daisy, what the project taught them about AI creativity, and where they see the technology heading next.

Where did the idea for Daisy come from?

Honestly, it started when we discovered and fell down the rabbit hole of scambaiting videos on YouTube. We were hooked on watching people like Jim Browning tie scammers up in knots, there’s something so satisfying about watching the scammers get scammed. But we also noticed the limitation: it wasn’t scalable. One scambaiter can only handle one call at a time. That’s where the idea clicked. What if we could take that same spirit and make it infinite and create the ultimate wolf in sheep’s cardigan? An AI granny who never gets tired, never needs sleep, and has all the time in the world.

Was there a particular decision, breakthrough, or risk that ultimately shaped the project’s success?

Yes, through testing, we saw that individual AI tools (speech-to-text, LLMs, voice synthesis) were just reaching the point where you could chain them together fast enough to actually fool someone over the phone. Latency was always the biggest hurdle; a few hundred milliseconds too long in Daisy’s response and the illusion breaks. It’s actually one of the reasons we cast an elderly protagonist. By leaning into the scammers’ own unconscious bias toward targeting the elderly, slower, more hesitant speech patterns, we gave Daisy a bit more breathing room while still feeling completely natural.

What’s one lesson from this project that you’ll carry into your future work?

Aim forward. Not all the technology behind Daisy was fully baked when we started, but based on our testing, we knew it was only a matter of time before it would be. By building ahead of where the tech officially stood, we got Daisy out into the world before anyone else could play in that space. That’s the lesson, don’t wait for the tools to be perfect, build for where they’re heading, especially if you have a client that can support that, like we did with O2.

Europe is home to a rich diversity of creative and technological perspectives. How does the culture of your region influence the way you approach digital work?

There’s a very British sensibility running through this whole campaign: dry, self-deprecating humour, a soft spot for eccentric old ladies, and a healthy distrust of anything that takes itself too seriously (including AI). That cultural instinct shaped Daisy entirely: rather than a slick, futuristic AI character, we built a slightly dotty granny with a cat and a pie recipe. It’s a very British way of disarming a scary technology, make it funny, make it human, make it a bit silly.

What trends, technologies, or shifts in internet culture are you paying closest attention to right now?

Building Daisy taught us a huge amount about AI personality design: how to give an AI character real texture, consistency, and a recognisable voice rather than generic chat responses. That’s set us up perfectly for the next big challenge brands are facing: making their AI-powered customer service agents feel genuinely brand-distinct, rather than the same cookie-cutter assistant everyone else is shipping.

What kind of work would you like to see more of on the Internet in the years ahead?

Less “now AI can replace this” and more genuinely creative applications that positively impact people’s lives. AI can’t just be about faster and cheaper, that undervalues what it’s actually capable of. It can also make things better, kinder, funnier, more protective, sometimes even more human. We’d love to see more work that treats that as ambition, not just an efficiency play.

If you could share one piece of advice with someone thinking of entering the Lovies, what would it be?

Respect the judges’ time. They’re busy, and they’re seeing hundreds of entries, so make your entry as easy to understand as possible. Don’t assume anyone’s heard of your campaign before. Keep the language engaging but not overly complex, since plenty of judges won’t have English as a first language. And try to hook them in the first ten seconds, that’s all the time you get to grab their attention.

Submit Your Best Work to Set the Benchmark for Digital Excellence in Europe.

If you are creating great work online, this is the moment to benchmark it. Explore the new and expanded AI & Innovation categories and measure your work against Europe’s leading voices. Enter before the Extended Entry Deadline on Friday, 7 August.