Last night, Intercom hosted an inspiring set of talks on AI in Product Design, showcasing experiments from teams at Figma, Monzo Bank, Intercom and more.

A few nuggets I took away:

1. Build your own AI design tools.
   Domingo didn’t explicitly say this, but his examples did. He showed how he used AI to gather every banner and notification across Intercom’s product so he could see them side by side. He then used AI again to build a small UI for fine-tuning micro-interactions like loading animations, shadows, and stacking behaviour, polishing the details live in the browser. Very cool.

   
   

2. Work small.
   Daria talked about carving off small increments of design to prototype with. (And ignoring the hype-filled promise of single prompts to build entire apps.) Good candidates are your riskiest assumptions, or the design details that will make the biggest difference to the experience. Think button interactions, dropdown behaviours, panel transitions.

   
   

3. Fifteen minutes a day goes a long way.
   Heldiney suggested setting aside a week where you spend 15 minutes a day building and sharing something with AI. That small, consistent habit compounds quickly and helps build confidence exploring AI tools. Something that echoes my experience with Hack Weeks participants. He also shared my favourite quote of the night: “If it looks bad, blame AI.” A good reminder that permission and psychological safety are key to creative experimentation.

   
   

4. Lovable and Cursor both have their place.
   I’ve often seen tools like Lovable, v0, and Replit as fun for demos but tricky for serious work. Cursor felt more “real.” Last night shifted that view a little. While Cursor suits production-level work, Lovable (or your vibe-coding platform of choice) shines when you need to make ideas tangible fast and move the conversation forward. Rather than focusing on what these tools lack, consider them when speed and accessibility is your top priority.

   
   

5. Figma still plays a role, but motion and behaviour matter more.
   In Figma, everything lives neatly inside static rectangles. But real experiences move, react, and sometimes even surprise us. As James shared, designing for agentic behaviour—where interactions unfold probabilistically—demands an emphasis on prototypes to fully communicate how experiences might unfold. This aligns with a neat tip from Daria about dropping code samples from codepen or easing.dev into your AI design projects to deliver more precise direction on polish.

But more than anything, I left feeling optimistic and energised. AI feels like it’s breathing life into London’s product design community, opening up new possibilities, new ways of working, and new territory to explore together.

Big thanks to the Intercom team for an inspiring evening. Plus the new faces, and old friends I got to catch up with: Dominic, Simon, Aarti, Florence, Domingo, Heldiney, Caitlin, Thom & Chris ❤️