Nothing Wants You to Build Apps by Talking to Your Phone

Nothing (the phone company, not the concept) just announced you can build apps by talking to your phone. Carl Pei's latest attempt to make Nothing phones worth buying involves AI that supposedly turns conversations into functional widgets.

"Vibe Coding" Is the Dumbest Name Ever

They actually call it "vibe coding." I'm not kidding. You tell your phone "make me a coffee tracker" and it allegedly builds you a working app.

The demo videos look slick - PDF generators, meeting prep tools, custom schedulers. But every AI demo looks perfect until you try to use it for anything beyond the exact scenario they showed you.

Community App Store for AI-Generated Trash

They're calling it the "Essential Apps Playground" where users can share their AI-built widgets. So basically an app store full of mini-apps that people created by talking to their phones.

This sounds like a recipe for disaster. Community-generated mobile content usually sucks. Remember when everyone thought user-generated games would change everything? We got thousands of broken clones and maybe three decent games.

Nothing claims they have "hundreds" of community tools from testing. Hundreds of what exactly? Working apps or half-broken demos that crash when you try to use them?

Local AI Processing = Dead Battery

Everything supposedly happens on your phone instead of the cloud, which is great for privacy but terrible for your battery life. Running AI powerful enough to generate functional apps locally sounds like a guaranteed way to kill your phone by lunch.

They built this on top of their screenshot organizer thing, which works fine for sorting images. But organizing files is way simpler than generating working code.

The Limits They Don't Advertise

The Phone 3 can only run six of these Essential apps at once. Older Nothing phones? Just two. That's not exactly screaming "revolutionary platform."

If your flagship phone can barely handle six AI-generated widgets without choking, maybe pump the brakes on the world-changing rhetoric.

Carl Pei's Marketing Playbook

Nothing keeps talking about "democratizing" app development and fighting "elitist systems." Dude, you're selling $400 phones with flashy lights. You're not exactly Gandhi here.

They love positioning themselves as the scrappy underdog taking on Apple and Google. Which would be more convincing if they sold phones to anyone outside of tech YouTubers and people who think transparent backs are the height of innovation.

Will This Kill App Developers? Nah

This won't disrupt shit. These are glorified widgets, not actual apps. Most people can't explain what they want clearly enough for humans to build it, let alone an AI that thinks every request needs a login screen.

I've been in meetings where non-technical people try to describe what they want built. It's always "make it like Instagram but for my specific hobby, but simpler, but also more features." Then they get mad when you tell them their idea doesn't make sense.

Real apps break in weird ways. Users do shit you never expected. Edge cases multiply like rabbits. You need developers who can say "no, that's impossible" and suggest alternatives that might actually work.

Security: The Ticking Time Bomb

Nothing says security "will be important" as they scale. That's startup speak for "we'll figure it out after someone gets hacked." AI-generated code is famous for security holes, and they want to put this on millions of phones.

Plus they're letting random people share AI-generated widgets. What could go wrong? How long before someone uploads a "weather app" that steals your photos or a "calculator" that crashes your phone every time you open it?

I give it six months before the first major security incident. "Oops, our AI accidentally generated malware" is going to be a fun headline.

The Long-Term Vision Problem

Pei talks about eventually supporting full-screen apps and voice-based creation. That's a nice vision, but Nothing needs to prove they can make the basic widget version work first.

The company has a habit of announcing ambitious features that sound cool in press releases but turn out half-baked in practice. Their track record on software execution isn't exactly stellar.

Market Share Reality

Nothing phones have tiny market share outside of tech enthusiast circles. Even if Essential Apps work perfectly, they're building a platform for a user base that barely exists.

The whole strategy feels backwards: build the revolutionary platform first, then worry about getting people to buy your phones. Usually it works the other way around.

Bottom Line

This could be cool if it actually works. But Nothing has a track record of overpromising and underdelivering on software features. The concept sounds great until you remember they can barely get basic Android updates out consistently.

If you have a Nothing phone, you'll probably mess around with this for a day or two. Whether you'll still be using it a month later depends on how well it works when you're not following their carefully crafted demo script.

Based on Nothing's software history, keep expectations in check.

Look, I know developers who've been building mobile apps for years and they're all tired of clients asking if ChatGPT can just "make them an app." Now Nothing is saying it can actually do that, which either means those developers are about to be unemployed or this is another overhyped demo that works great until you try to build anything beyond a to-do list.

The App Development Reality Check

Here's what nobody talks about: most custom apps cost $50k-200k because clients have no fucking idea what they actually want. They say "like Uber but for dog walkers" and six months later you're building a calendar app with photo uploads because that's what they really needed but couldn't articulate.

Nothing's betting that AI can understand user requests better than the users themselves understand their own needs. Having sat through hundreds of client meetings where "simple" means "incredibly complex but I don't want to pay for it," I'm skeptical.

The widget limitation is smart though - easier to nail "show me the weather" than "build me a social network that doesn't suck."

How Nothing Plans to Make Money (Eventually)

Right now Nothing is giving this away for free, which means they're either playing the long game or haven't figured out monetization yet. My money's on "haven't figured it out."

App stores take their 30% cut because they handle payment processing, hosting, review, and distribution. Nothing's community model skips all that overhead, but also skips the revenue. Someone still needs to pay for the servers running the AI that generates these apps.

They'll probably try the usual startup playbook: get users hooked, then add premium features. Maybe charge for advanced AI models, or take a cut when people share popular templates. But first they need to prove people actually want this beyond the initial "cool demo" phase.

What This Means for Actual Developers

Will AI widgets kill app development? Not fucking likely. These are glorified shortcuts, not real apps. I've seen the demo where you say "make me a tip calculator" and it works. Cool. Now ask it to handle edge cases like split bills, tax variations, or custom tip percentages and watch it fall apart.

Real apps have to handle thousands of edge cases that users never think about until they hit them. Authentication flows, data persistence, offline functionality, accessibility compliance - AI demos never show you the boring shit that takes 80% of development time.

If anything, this might create more work for developers. Someone needs to build the backend APIs these AI widgets will connect to. Someone needs to debug when the AI generates code that crashes on Android 12 but works fine on Android 13.

The Market Share Problem

Nothing has less than 1% of smartphone market share, which makes their platform strategy feel backwards. Usually you get people to use your phones first, then you can push software features. They're trying to build a revolutionary platform for the twelve people who actually bought Nothing phones.

Apple can force developers to use their tools because everyone has iPhones. Google can push Android features because Android is everywhere. Nothing can... hope that their AI widgets are so amazing that people switch phones? Good luck with that.

This feels like another case of a startup believing their own press releases. Yes, the technology is cool. No, cool technology doesn't automatically mean market adoption.

The Technical Reality No One Talks About

Running AI models locally on phones is a battery-killing nightmare. I've tested apps that use on-device ML and they drain power faster than Bitcoin mining. Nothing claims their AI runs locally for privacy, but that means every app generation is competing with your normal phone usage for CPU cycles.

The Phone 3 supposedly handles six Essential apps. Six! My production iPhone runs 50+ apps without breaking a sweat. If your revolutionary platform maxes out at six widgets, maybe the platform isn't ready for prime time.

They'll need cloud processing eventually, which kills their privacy angle. Then they're just another company harvesting user data to improve their AI models.

The Privacy vs. Performance Trade-off

Nothing's "everything local" privacy approach sounds great until you realize local AI models suck compared to cloud ones. GPT-4 works because it has massive server farms behind it. The AI model that fits on your phone? It's like comparing a Ferrari to a golf cart.

Everyone wants privacy until they want their apps to actually work well. Look at Siri versus ChatGPT - one runs locally and barely understands you, the other needs cloud processing but actually helps.

Nothing's betting that people will choose worse functionality for better privacy. Maybe they're right, but most users choose convenience over privacy every time.

The Chicken and Egg Problem

Nothing needs people to create and share AI-generated apps to make the platform valuable. But people won't create apps unless there's already a community using them. Classic startup problem: you need users to attract users.

Their plan seems to be "build it and they will come," which worked for Field of Dreams but rarely works for tech platforms. Even with their existing Nothing phone users, that's still a tiny base to build from.

The open-source angle could help - other phone makers could add Essential support. But then Nothing loses their main differentiator and becomes just another Android customization.

Following the Money

Nothing raised $200 million for this, which gives them maybe 2-3 years to prove it works before investors start asking uncomfortable questions about revenue. That's not much time to build a platform, grow a community, and figure out how to make money.

The metrics that matter aren't complex business school bullshit - it's simple: do people actually use these AI-generated apps after the novelty wears off? And are they willing to pay for better ones?

My guess is they'll burn through most of that funding before they figure out sustainable revenue. Then they'll either get acquired by a bigger company or shut down, depending on how much attention they get.

The Security Nightmare Waiting to Happen

Nobody's talking about the security implications here. AI-generated code is notoriously buggy and full of vulnerabilities. Now Nothing wants to let random users share AI-generated apps with each other? What could possibly go wrong?

When (not if) someone uploads a "weather app" that steals your photos or a "calculator" that crashes your phone, who's responsible? Apple reviews every app manually before it hits their store. Nothing's plan seems to be "let the AI figure it out."

This feels like the early days of Android when malware was everywhere because Google didn't vet anything properly. Except now it's harder to detect because the code was generated by AI instead of written by obvious scammers.

What Happens If This Actually Works?

If Nothing somehow pulls this off - if AI-generated apps actually become useful and popular - then Apple and Google are fucked. Their entire app store model depends on controlling development tools and distribution.

But I doubt we're there yet. This feels like 2008 when everyone thought Flash on mobile was the future, or 2012 when everyone thought HTML5 apps would replace native apps. Cool demos, but reality is messier.

Apple will probably add some AI coding assistance to Xcode eventually, but they're not going to let users generate and share unvetted apps. Google might be more open to it, since they already allow sideloading.

The more likely outcome? Nothing gets some buzz, investors get excited, then reality sets in when people realize AI-generated widgets aren't actually that useful. The company pivots to "AI-powered development tools for professionals" and gets acquired by Microsoft or someone for their talent.