AI with Kyle Daily Update 122

Today in AI: Apple gives up

What’s happening in the world of AI:

Highlights

🍎 Apple Surrenders: Siri Will Run on Google Gemini

Apple and Google have entered into a multi-year collaboration. The next generation of Apple's foundation models will be based on Google's Gemini. These models will power future Apple Intelligence features, including a more personalised Siri coming this year.

Steve Jobs would be rolling in his grave.

This is an absolute affront to everything Apple has stood for - closed ecosystem, quality control, building everything in-house within their walled garden. No more. They're handing Google the keys to the kingdom.

For years, Apple fans kept saying "just wait, they're working on something big." They did this with the iPod, they did it with the iPhone. They waited until the market matured and then POW - released the best version.

I was always suspicious though - you can't develop a foundation-level AI model in secret as a public company. We'd see the data centre costs and GPU investments ramping through the roof. We didn't.

Then they released Apple Intelligence - not good. AirPod Pro translation - used it once, terrible. Genmoji - lackluster. This announcement is the final nail in the coffin.

They have nothing.

The head of Siri has been collecting tens of millions for 13 years and now has to ask Google to fix it. Embarrassing.

Kyle's take: The consumer spin is privacy - on-device processing, Private Cloud Compute, all that. But do people actually care? Nah.

ChatGPT has 900 million users and most don't give a flying fiddlestick (I managed not. to swear on the livestream…) about privacy. The bigger risk is the Microsoft Copilot problem - Gemini underneath but neutered with so many guardrails it ends up rubbish anyway.

I’m an Apple fanboi but I’m not impressed.

🛠️ Claude Launches Cowork - Claude Code for Everyone

Claude Code has been the best coding agent on the market, but the name puts people off. It has "code" in it. The interface is intimidating. That just changed.

Cowork lets you give Claude access to a folder on your computer. It can read, edit, and create files. Try it to create a spreadsheet from screenshots or a draft from scattered notes. You describe what you want, it makes a plan, writes notes to itself, then works steadily through the task. It asks before taking significant actions so you can course correct.

Currently available as a research preview for Claude Max subscribers ($200/month) in the macOS app. Others can join the waitlist. Here’s the waitlist link : https://forms.gle/mtoJrd8kfYny29jQ9 

Kyle's take: The positioning is odd - at $200/month you're probably already technical enough to use Claude Code. They need to bring this to the $20 tier for it to have seismic impact.

Here’s the BIG thing though. Cowork was built in one and a half weeks. Claude Code wrote 100% of the code. Claude Code created a simpler version of itself and Anthropic shipped it. The world is changing very quickly.

😬 Bad Week for Manus

Manus is in the middle of a multi-billion dollar acquisition by Meta. Cowork looks an awful lot like what they do - agentic system, helps with research, takes control of your computer. The difference: Anthropic built it in a week and a half.

Adding to their troubles, Beijing is now reviewing the Meta-Manus deal. Manus was founded in China (headquartered in Singapore for regulatory reasons). First time China has really tightened control over AI company exports.

Kyle's take: That acquisition suddenly looks less attractive when something similar can be made in a week. Sorry, a week and a half. The Manus team has built something cool but they're probably having a rough week.

🌍 How to Actually Prepare for AI (David Shapiro Framework)

David Shapiro released a video on “high-agency moves” for AI disruption. Worth watching in full, but the highlights:

Consider relocating - if corporate jobs disappear, where would you want to live? I'm moving from London to Cyprus this year. Going from surviving to actually saving money. Build a financial buffer - if you make £80K, act like you make £60K. We're about to enter chaos and any assets will cushion the blow.

Four job categories likely to stick around: The attention economy (content creation driving business, not just ad revenue). The experience economy (in-person, human presence stuff). The authenticity economy (face on camera, skin in the game - AI will demolish anything faceless). The meaning economy (helping people make sense of identity loss when AI takes their profession).

Member Question from Lyra: "Any suggestions when editing in Lovable and something changes I didn't ask for and it breaks?"

Kyle's response: Welcome to vibe coding. First, connect to GitHub - icon in top right, it's free. Every change saves a version you can roll back to. When something breaks: right-click in browser, click Inspect, go to Console. Copy the red error messages and paste them back to Lovable saying "I just got this error, please fix it." Usually sorts it out.

More advanced? Ask for multiple potential solutions. This forces your tool to “think” through a problem rather than just rush to solve it. Works a treat.

Or…move to Claude Code. It’s amazing how many problems Claude Code can just one-shot that may have caused weeks of pain previously!

Want the full unfiltered discussion? Join me tomorrow for the daily AI news live stream where we dig into the stories and you can ask questions directly.

Streaming on YouTube (with screen share) and TikTok (follow and turn on notifications for Live Notification).

Audio Podcast on iTunes and Spotify.