Feels like we’re still very early in figuring out what these devices should actually become.
New hardware isn’t interesting just because intelligence has been moved into a new shape. The more interesting question is what new interaction patterns become possible once a computer can perceive what’s happening around it, remember context, infer intent and act on its own.
And then which of those interactions only make sense because the intelligence is embodied in a particular device, with a particular set of sensors, a particular physical presence and a particular relationship to the person using it.
We played with some of this quite a bit @nothing, and kept coming back to starting with the interaction, then figuring out what kind of device it actually wanted to live in. The form factor usually became a lot clearer once we knew the interaction was actually useful.
Going through the Braun archive chronologically and seeing the design language evolve was incredibly inspiring.
One thing I hadn’t realised was how the shaver business was basically the reliable part of the business, giving them room to take much bigger risks with audio.
There’s something really nice about a reliable business quietly funding the experimental stuff. We need more companies set up like this.
Calling all London design people! 🇬🇧
There is an awesome collection of vintage Braun products designed by Dieter Rams in Shoreditch
Sadly the owner has to move out at the end of September, but I was able to get him to put on 2 more events:
📍Tonight - 6-9pm - Sold out 🎟️🎟️
📍Thursday 24th 6-9pm - tickets available
If you're interested, the luma link is below!
Speakers at The Cause in London. Basically just massive acoustic horns
So utilitarian and somehow still such cool objects
I can’t wait to be able to control my phone completely hands-free. I keep hitting things where I have to pick it up and unlock it just to do something basic like changing a song or running a search.
Speaker recognition already seems pretty good. Spoofing is obviously the harder bit, but it feels like we could work around some of that with a continuously updated auth state combining voice, nearby devices, recent phone unlocks and maybe other sensor signals.
Makes me wonder what’s actually blocking this now?
Been using Grok Bot and Instinct a bunch and realised I don’t really care about seeing more reasoning. I care about being able to see what the agent is actually doing, especially when it’s using my accounts.
With Instinct I mostly send a message and hope for the best...
Gemini 3.8 Live is the first voice model I've tried where I can speak Lebanese naturally: Arabic, French and English mixed together within the same sentence.
It still falls back on the same word choices a little too often, but the interesting part is that I’ve mostly stopped adapting how I speak to the model.
That feels like a meaningful threshold for voice interfaces.
Walid Behlock retweeted
Big day. Two new products from @nothing: Phone (4b) and Ear (3a).
Our team has balanced smarter tech and bolder personality with a more accessible price point. I’m so proud of what they’ve achieved and already enjoying the early reaction.
“Something about Nothing still feels different in a market where most smartphones have become variations of the same idea” - @CreativeBloq
“Nothing's entry-level tech remains very impressive from a value point of view” - @T3dotcom
“Nothing peps up its range with what looks like another winner.” - @wallpapermag
…and our RCB edition of the Phone (4b) sold out in India in under two hours.
We believe these products show new tech doesn’t have to be boring. Thanks to everyone that’s been part of the journey so far. Much more to come.
glyphs + sounds + haptics
voice + tone + pacing
we're going for a point of view, an aesthetic, coherent across every surface
The MCP backlash today has been interesting to witness... turns out models are capable enough to read the messy web as is and figure it out
The Semantic Web tried to impose standards to make data machine readable 25 years ago, never hit critical mass. Same thing will happen with most of our tools, agents will learn to use them from docs and playing around, without the need for a universal spec
This is especially true on mobile, where agents that can navigate messy UIs and fragmented data will beat ones waiting for every niche app developer to implement a spec. Samsung is already betting on this with their S26, where Gemini navigates real app UIs in the background instead of asking developers for APIs. Right direction, but fragile
The real solve will be an OS level agent that understands you deeply enough that it doesn't need to pretend to be a user. Apps just need to expose a handful of core actions (order food, book ride, play playlist) as structured intents and let a lightweight on-device agent figure it out. Android is well positioned here, the foundation already exists
The blind box model (Labubu, gashapon) is centred around the transaction, that's when the dopamine hit happens. You either love it, trade it, or buy another box chasing rarity. The cultural power is also external, it's signalling, belonging, social currency.
STARBOY is interesting as it inverts both mechanics. The randomness doesn't happen at purchase, it evolves over the lifetime of the object, creating some kind of relationship model. You can't grind for a better outcome, can't optimise it. That's a different reward loop to anything in consumer tech right now.
Pretty sick release
Walid Behlock retweeted
The new Nothing 4a Pro is here!
This keeps proving that Android has a wider advantage in uniqueness, personalisation and style.
And better cameras for less!
Walid Behlock retweeted
This isn’t the first time a shift like this happens and it won’t be the last.
I feel like what’s happening right now is what happened to photography — when everyone has an amazing camera and a publishing outlet to the entire world, we just get more of everything. Diversity and new ideas, class differences erased. But also a huge noise increase.
The people who would wade through the mud of challenges to master photography pre iPhones were on a path of commitment to being the best they could be. Nowadays the vast ocean of noise has weirdly dampened photography as an art. I for sure thought the opposite would happen, that people without the means to pursue photography in the pre-iPhone times would now have a stage, but they too drowned in the noise.
The same thing happened to sign making and many other fields of craft. It’s happening to software now. Average actual quality will go down and consumer expectations on quality will fall. Maybe the next 10-15 years is simply the cost of progress and we’ll build Star Trek-grade computers in a few generations from now.
Perhaps this is “good”, perhaps not. Regardless, it feels like desperate times of gold rush, not like a renaissance of software.
In some ways, software around 2010 felt like peak avg of quality. What did we culturally do back then that we changed or stopped doing, I wonder…
Bots will have their own digital identity, agency, and communication methods
We're rapidly moving past "bots that use human interfaces" toward native agent-to-agent protocols
What gets unlocked when agents can discover, authenticate, and transact with each other without human readable intermediaries?
Bots as first-class internet citizens
