💯
Posts by Toby Rogers
Book about exploring artificial intelligence on your Commodore 64.
Yeah, that’s always a risk. Used well, though, visual prototypes can spark that “is this the right thing?” conversation in ways that abstract lo-fi stuff doesn’t, especially at that really early strategy stage. You need to be explicit about their use as tools for thinking, not solutions, though.
One of the best use cases I’ve found for AI prototyping is rapid concept generation.
We ran a workshop with a bunch of c-suite types using Figma Make to create visualizations of propositions on the fly—not to get to a solution, but to give us tangible ideas to talk around.
It worked really well.
Look at me doing a podcast 🎙️
No experience you design will ever delight your users more than a pomme bear
Missed this yesterday but still worth sharing 😆
We’re all Rick Rubin now
AI CEOs when asked whose jobs they plan to replace next:
“business idiot” is my favorite descriptor/euphemism of a certain type of MBA class of person.
the "what you made with sora mattered" is living rent free in my head as a *very important statement*
Its absurdity or ominousness lies in the definition "mattered"
The absurd definition, where nothing made with sora actually mattered, is applicable to *anything* made with generative AI
This is super-creepy and concerning, but not in the least bit surprising.
Can’t decide if Sora shutting down is the first signal that the AI bubble is bursting, or the first step towards LLMs becoming boring infrastructure instead of media slop machines.
Sending your CEO a strategy deck an LLM made for you is like getting up on stage to mime a song you’ve never learned to play.
Ninety percent of new product ideas are shit but now we get to experience all of them in their vibe-coded glory instead of staying where they belong in someone’s “best product ideas in the world” Apple note.
AI is getting really good at completing tasks, but it’s still terrible at taking over complete jobs.
The AI is replacing workers narrative only works for simple, low risk, transactional stuff.
Anything that requires judgement, experience and understanding still needs humans.
After all the “I’ll use an agent to book my flights” demos, how many people are actually letting agents just book their flights?
This is the fundamental truth about AI right now. If you're good at your job, AI can make you several magnitudes better. If you're bad at your job, though, it'll make you even worse (to the point where the robots might as well be doing it instead).
This is interesting. The user experience for agentic shopping sounds like it's going to be "let an LLM take me to the store" not "let AI do my shopping for me" (for now, at least).
If an AI can do it, is it worth doing?
The Occam’s Razor of BS Jobs
As a PM, you should be using AI to free yourself from busywork so you've got more time to think, not letting it do the thinking for you.
When everyone has the same superpower, it’s not really a superpower.
All I'm going to say that: If I had never done a thing, I wouldn't be so quick to jump into discussions about whether "AI" can do the thing.
One quiet superpower in product management:
Recognizing the real decision before anyone else does.
But if you can see the layer the debate is happening in, you can help the team stop arguing and start deciding.
That’s often more valuable than having the answer.
Figma stock dropped -8% today as Google introduced “vibe designing” into its Stitch UI tool which lets you describe the UI you want to create.
Figma is now down -80% since its IPO which is just brutal. I never have expected them to be one of the early casualties of AI.
Snow Crash wasn’t meant to be aspirational
If you think about it, the cost of building software has always been close to zero.
You’ve never needed much more than a computer and a text editor.
Apps like Lovable, Replit and Bolt don’t really change that.
What they change is who can build, the same way GarageBand changed who can make music.