The new tech is busy but not useful. "intro": "Meta's Muse watches you. Siri got a facelift and nobody noticed. Agents are the new app — they do the work. TradeWork does one thing for the trades. The GSA office smells like wet drywall and has mice. Why this matters for us: we're paying f
Meta's Muse is better at watching you than helping you
Meta's Muse is the new app that keeps the company's favorite habit: it collects your data for AI training whether you agree or not. But this one goes further. It nudges you to share your bank account, your email, and even your passport. That last one is the real tell — Meta wants your passport data, and it's asking nicely.
The app is supposed to be helpful. Instead, it's more interested in surveillance than service. Every prompt is a chance to hand over another piece of your life for the AI engine.
Why this matters for us: if Meta's building its AI on our bank accounts and passports, it's quietly building a profile on Brown families the way la migra builds one on us — through data we hand over without really noticing.
Siri got a facelift, then I forgot it existed
Apple refreshed its phone assistant this summer. For a few months, it was the new shiny thing — a brief summer fling with Siri AI.
Now the full release is creeping closer and the writer has already moved on. The honeymoon is over. You don't need to install anything to notice this one: it's the built-in assistant you already have, the one that lives in the Control Center and the lock screen.
The story is simple. Apple gave it new looks and new tricks. People liked it. Then life happened, and they stopped thinking about it. That's how most tech lands — a wave of attention, then the quiet.
No dramatic launch. No keynote speech. Just a quiet upgrade to something you already use every day.
Why this matters for us: the assistants that survive aren't the ones with the flashiest launch — they're the ones that become background noise, the ones we stop noticing because they finally do what they're supposed to do.
Agents are the new app. They do the work, not just show it.
Agents are AI programs that do things instead of just answering questions.
A regular chatbot is like a friend who listens to your problems. You tell them what's wrong, they reply. That's useful, but it stops at the conversation.
An agent is more like a cousin who actually handles the errand. You say "send the invoice to Mrs. Lopez" and the agent opens your email app, types the message, and sends it. It doesn't just tell you how to do it. It does it.
The difference matters because agents connect to real tools — your calendar, your bank, your drive, your CRM. They can read your files, make decisions, and take action across apps without you clicking through each one.
Think of it like the difference between a menu and a waiter. The menu tells you what's available. The waiter brings you what you ordered. Agents are the waiter.
This is how the tech world is moving: from apps that sit on your screen to agents that work behind it. You'll see more of this in the next year.
If you're using AI and still doing all the clicking, you're using the wrong tool. Ask what it can actually do, not just what it can say.
Our cameras stay put — the question is who's steering them now.
— theverge.com
#gopro-s-boss-says-the-cameras-stay-even-after-the-285m-buyout-cb40d4Robotaxis are the new villains — a short film has them on the loose
Road Rage is a short film about a San Francisco car chase where the other driver isn't human and might be trying to kill you. The premise is simple enough, but what matters is the cultural mood behind it. A robotaxi can now be the bad guy without any explanation at all.
The film is on Kickstarter for $40,000. It follows Joa and leans into the same anxieties that keep surfacing: autonomous cars have gone from sci-fi curiosity to street-level reality in a handful of American cities. They're stand-ins now for bigger worries — AI eating jobs, Big Tech watching, machines making choices that used to belong to people.
TradeWork — one app for the trades
The trades are still running on notes apps, PDF invoices, text threads, and checks. It works—barely. Estimates get lost. Crews double-book. Payments sit in inboxes waiting to be matched. General-purpose CRMs are too heavy for a foreman juggling jobs. Trade-specific apps usually live in the office, not the truck.
TradeWork is built for the truck. Mobile-first. Jobs, crews, invoices, and payments in one place. Bilingual surfaces so the office, the foreman, and the helper all read the same job without translating it back and forth. An estimate becomes an invoice. An invoice becomes a payment. No extra clicks, no extra tabs.
Why this matters for us: it keeps the tradespeople who build and fix our neighborhoods from losing money to paperwork.
https://tradework.work
Gemini Spark now runs your Google Photos library
Google rolled out a new feature for Gemini Spark that lets it manage your photo stash without you touching the app. It edits albums, curates collections, creates shared albums, and even turns photos into calendar events. All of this works for Google Photos subscribers on the AI Pro and AI Ultra tiers.
The idea is simple: instead of scrolling through thousands of pictures, you can ask Gemini Spark to do the sorting for you. It figures out which photos belong together, which ones deserve a shared album for the familia, and which ones should become reminders on your calendar.
SpaceX makes its own turbine blades to power the AI data centers
SpaceX is manufacturing its own turbine blades in-house to feed the gas-powered generators for Elon's AI data centers. The move cuts generator delivery delays by 18 months.
It sounds like a small part, but the blades are what make a turbine spin — the heart of a gas generator. Without them, the whole unit waits. By making them in-house, SpaceX sidesteps the supply chain bottleneck.
This is part of a bigger shift. The company is no longer just launching rockets. It's building the engines, the turbines, the parts — whatever it needs to keep the machines running. The gas generators will spin up the servers. The blades keep them turning.
For the tech world, this is another signal that the companies building AI at scale are pulling more of the supply chain inside. When demand spikes, you don't want to be at the mercy of a supplier who can't deliver.
Why this matters for us: the same supply chain tightrope that holds up a turbine blade also holds up the parts we need for our shops, our generators, our trucks — and when the big players pull manufacturing in-house, it changes the market for the rest of us.
The GSA's new office smells like wet drywall — and has mice
The General Services Administration just moved into a shiny new building in Washington. The lease is steep — $48 million a year for 1.7 million square feet. The rent is higher than the old building, but the old building was 70 years old. The new one should be better. It isn't.
Workers are finding mice in the ceiling voids — the same ones the old building had. The drywall is sagging and caving in. The water tastes cloudy and has a chemical edge. People are asking whether it's safe to drink. One worker told Wired she looks at the ceiling before going to the bathroom.
Run a chatbot on your own machine — no cloud, no privacy tax
A local LLM lives on your laptop. You run it yourself. No API calls. No company's servers reading your prompts.
The idea is simple: download a model, install the software, and talk to it right there. Your questions stay on your machine. Your data doesn't get sold or logged.
For most people this means a 7B–13B parameter model. It's fast enough for daily use. It handles email, notes, code — the stuff you actually do. You point it at your files and it reads them without uploading anything.
If your machine is beefy, you can go bigger. If it's modest, the smaller models are plenty. The trade-off is size versus speed. You pick what works for your hardware and move on.
This isn't a niche thing for engineers. It's for anyone who doesn't want their queries floating off to some data center, where someone — or something — might be listening.
Why this matters for us: the gente who run side businesses, freelancers, and families who share laptops need tools that don't siphon their data to the cloud every time they ask a question.
The iPhone Fold is coming — and the concert-view-blocking is about to get worse
Apple announced its folding phone this week. The hinge is the tricky part: a crease, some weight, a thicker body. It works. It's just not the kind of gadget the abuela would actually buy.
But the real story is what it means for the rest of the ecosystem. The unfolded screen is twice as wide. For recording video — the way our tías and la migra app users do it — that's a wider frame but a taller, more awkward hold. You're now the person in the back of the concert, phone out, blocking everyone behind you. The foldable is just the big version of the same problem.
Building an AI factory on Kubernetes
Cloudflare's new post walks through how they keep DNS cache from gobbling up RAM — a real problem when you're running thousands of zones and the cache grows until it chokes the rest of the system. The fix is a smarter eviction policy that bumps the least useful entries first instead of just dumping the oldest ones.
The same week, the CNCF published a how-to on standing up an AI inference stack on Kubernetes. They're treating GPU nodes like any other resource, scheduling models and workers across the cluster so nothing sits idle. The key insight is that AI workloads aren't fundamentally different from the apps we've been running on k8s for years — they just need more memory and faster GPUs.
Both posts are for the people who actually run the infrastructure: the engineers who stay up when something breaks, not the VCs pitching the next platform. They're practical, not promotional.
Why this matters for us: the same k8s clusters running our apps are now handling AI workloads too — and the cache bugs that slow down DNS also slow down inference, so the fixes here apply to the whole stack we depend on.