396 - Matt Webb

Matt Webb is the founder of Acts Not Facts, a vehicle for product invention.

Through that, he is also the inventor of Poem/1, an AI-powered rhyming clock. How does it work? It tells the time with a brand new poem every minute, composed by ChatGPT. It’s sometimes profound, and sometimes weird, and very occasionally it fibs about what the actual time is to make a rhyme work... It has a cute e-paper screen.

So far it's been featured in Fast Company, Forbes, Ars Technica, The Verge, Bloomberg, the New York Times and more.

You can back the kickstarter here. It's pretty neat!

Twitter (X) → twitter.com/genmon
Threads → threads.net/@genmon
Instagram → instagram.com/genmon
LinkedIn → linkedin.com/in/genmon

Inside Matt's Workspace

Workspace Items: Desk

• Mac mini M1
• MacBook Air M2 13” — this is the best laptop Apple have made since the 12” aluminium Powerbook
• I don’t have a KVM switch — I use the machines simultaneously using Apple’s Universal Control it is a robust as hell. I forget there are two computers there
• Ring light because of the pandemic
• Aranet4 CO2 monitor — if I forget to open a window and it gets about 800ppm I get nervous. Above 1000ppm and I start getting stupider
• Cricket ball as fidget device
• On the top shelves are Wisden cricket almanacs and a piece of metal I use to burn incense which is a part of the Concorde fuselage, a project that my parents were attached to

Workspace Items: Shelves

• The book situation is out of control. Visible sections are cybernetics/computing history, history, mythology, and design. Psychology and animals are off to the side
• In the middle: I leave this camera set up to snap any good poems from the Poem/1 technical prototype. Camera is a Sony NEX-5N. It’s old but there’s nothing like it for size, just perfect in the hand and pocket
• HomePod mini, the only smart speaker with an open mic in the house
• “Twinkly” brand Christmas lights. The app uses computer vision to place the RGB LEDs in 3D space, then you can run animated patterns over them
  
• Many ornaments and bits and bobs! There’s an old FM3 Buddha Machine, archive boxes for issues of JBIS (Journal of the British Planetary Society, which I throughly recommend) and an original LOT 2046 tattoo machine (as-yet unused…)
• The cushion on the arm chair is in the new Elizabeth line moquette

How do you spark creativity?

Pulling random books and browsing.

For instance: I recently took a deep dive into human/AI interactions and ended up creating a software sketch of AI NPCs on a multiplayer whiteboard. Here’s the writeup.

‍It was surprisingly hard to get into, and I couldn’t - to begin with - come to terms with the “voice” of the NPC. How smart should these virtual users be? Should they speak to you? Should they do many things, or one? Is it ok to give them instructions? The breakthrough was a piece of work by the architecture practice Ant Farm who designed a dolphin embassy in 1974. Dolphins are super smart like humans but definitely not human. Like AI, right? So to break the block, I started by sketching a dolphin embassy on my canvas. All my NPCs were dolphins. That gave me the constraint I needed to get to a working experience prototype, then I could kick away the metaphor scaffolding and design something that made sense instead. I’ve still got the old sketches around someone.

Why are you creating Poem/1?

That spot in the middle of the shelves? That’s where my old e-ink clock was, the one I put poems on to tell the time with ChatGPT. I was sat in the sofa off to the side (not seen) just laughing at it.

That’s enough, that’s what I look for in any work. Something that makes me feel something new. I don’t believe you can work your way towards a new experience. You unearth it, glimpse an edge of it. Then all the work is digging it up and polishing it and framing it and so on.

It is hilarious to have a clock that tells the time with a new poem every minute, every day, composed by ChatGPT. It is profound, weird, sometimes bizarrely appropriate… I end up over time with a really visceral intuition of what AI is good at and what it isn’t good at. It’s not intelligence, not at all, but it’s also not like the computers we’re used to, it’s a step change from that. So I enjoy that encounter.

But it makes me laugh. So I want a beautifully designed one on my bookshelves. And the most straightforward way to get one is to persuade 1,000 other people that they want one too.

What was the most challenging aspect of getting Poem/1 on Kickstarter?

I could say it’s the number of skills involved. I’m grateful that London has such an awesome hardware scene and I get to collaborate with folks like Approach Studio and Tom Armitage.

I could say that it’s the number of moving parts and that every part brings a new constraint: firmware, servers, component cost and availability, shape and size, power, markets and margins, the economics of sustainability, the right partners…

But really the most challenging aspect is attachment. My studio is Acts Not Facts and though I hope to grow it, it’s just begun. And bringing connected hardware into the world as a one person studio is very unlikely. It’s a long process, you can’t push it. You chip forward, chip forward, chip forward. Maybe it’ll work, maybe it won’t. In any long project, you develop emotional attachment. You want it. Yet at any given moment, it probably won’t happen.

How do you maintain looseness in the face of that want and fear? How do you keep soft hands and an open ear to possibilities and suggestions? It’s hard!

So it sounds counterintuitive but I worked hard to cultivate high passion, low attachment and sense of humour during the development of Poem/1. I think that comes across in the Kickstarter campaign! And if the campaign succeeds (I hope it does) I will be able to relax a little because the path forward will be clear.