Look at more

Hello from London,

Since moving back to London, I keep noticing the Deliveroo kangaroos. Every few minutes one goes past on the back of a bike, weaving through traffic in the rain. We shaped that brand in 2016. I still get a small jolt seeing it out there, on someone else's commute.

There's a conversation with Will Shu on the site about how we shaped and launched it.

Watch the conversation — Delivering Success →

I've also been looking at Taste Labs, which is trying to teach AI to make subjective judgements. Having spent a fair amount of my working life defending a typeface in a meeting, I'm curious about what it will learn to agree with.

Over in YC's automation directory, the ambitions are more specific. Bizmark describes an agent that handles procurement and quoting for factories. Infera turns a plain-English experiment into instructions for lab equipment. Both list just two people.

I can see the appeal of handing over the quoting. I'm less sure how I'd teach a machine why one typeface is right. Trying to explain it would probably be good for me.

Meanwhile, at work

She's back

Serena Williams trusted us with the secret of her comeback. We had two weeks to make sure Factor_ was standing next to her when she stepped back into the spotlight. This was the first launch in a multi-year partnership.

See She's back →

The Countdown

For The Countdown, Maximum Effort was back in the room and we had more of the HelloFresh team involved. It felt like we were hitting our stride together.

See The Countdown →

What the feeds taught me

The little vibe-coded sites have turned into a way of making myself pay attention. Type Foundry Letters keeps me looking at details; I keep returning to older pieces from Rogue School on learning faster and Assembly on AI taking over junior work. If we hand over the work, where do we get the practice? I hadn't expected that question to come out of setting up some research feeds. Andy Stefanovich would recognise the instruction: Look at More.

More soon,