AI with Kyle - Daily AI News and Updates

Free how-to-setup OpenClaw guide: https://aiwithkyle.com/openclaw

Get AI-Ready with Kyle’s 5-Day Challenge: https://aiwithkyle.com/join

Kyle’s AI Canon — Curated list of AI learning essentials: https://aiwithkyle.com/ai-canon

Subscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQg

This video is from the ‘AI with Kyle Live Stream’

First aired: 16 February 2026

Summary: OpenAI just acquired OpenClaw from Peter Steinberger, and honestly, the timeline on this is absolutely bonkers. We're talking 19 days between Anthropic sending him a cease-and-desist letter for calling it "Claude Bot" and him selling to OpenAI for what I imagine is around a billion dollars. He's one guy from Austria who's been grinding away on GitHub projects for years, and suddenly every major AI lab in San Francisco is chasing him down. Sam Altman got him, Zuckerberg tried, and Peter's now moving to San Francisco to build a version of OpenClaw that even his mum can use. The project's staying open source in a foundation, which is promising, though I suspect OpenAI will spin off a closed-source version where the real development happens.

The big losers here? Anthropic, who fumbled this spectacularly by only ever communicating with Peter through legal threats instead of actually talking to him. They could've partnered with the fastest-growing open source AI project of all time, but instead they're the baddies in this story. And Europe—yet again—loses top talent to Silicon Valley because we simply don't have the ecosystem or resources to compete. Fair play to Peter though. Thirteen years of building, 40+ projects on GitHub, and he's finally hit something genuinely transformative. OpenClaw is essentially the Jarvis we were promised—a persistent AI assistant that actually works proactively rather than just sitting there waiting to be asked. It's got rough edges and security concerns, but as a preview of what's coming, it's bloody exciting.

—— Time Stamps ——

00:00 OpenClaw Sold to OpenAI: How One Developer Became a Billionaire

—— More Useful Resources ——

Find everything else at https://aiwithkyle.com/

Get AI-Ready with Kyle’s 5-Day Challenge: https://aiwithkyle.com/join

Kyle’s AI Canon — Curated list of AI learning essentials: https://aiwithkyle.com/ai-canon

Subscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQg

This video is from the ‘AI with Kyle Live Stream’

First aired: 13 February 2026

Summary: Friday the 13th and I'm walking through something I've been chatting about loads on my socials lately—how to actually validate a business idea before you waste months building something nobody wants. It's called Lean Testing, borrowed from the Lean Startup methodology, and with AI we can now test ideas ridiculously quickly. The concept is beautifully simple: flip the entire process around. Instead of building in secret for six months, polishing everything to perfection, then launching to crickets, you sell the thing first and build it after you know people actually want it. Sounds backwards, feels backwards, but it works.

I walked through my actual setup from yesterday—the sales pages, the Meta ads, the whole flow—showing how you can test multiple variations of an idea without touching a line of code or building the actual product. The beautiful part is you avoid that soul-crushing moment where you've poured your heart into something and the market just shrugs. I also got into some chat about the current AI coding landscape—Codex 5.3 is genuinely brilliant, first time I've seen something match Claude Code—and Google's Chrome AI assistant which is quietly positioning them rather well in this race. But the core message: talk to your market before you build anything. Revolutionary concept, apparently.

—— Time Stamps ——

0:00 What is Lean Testing? Validate Business Ideas with AI

—— More Useful Resources ——

Find everything else at https://aiwithkyle.com/

Get AI-Ready with Kyle’s 5-Day Challenge: https://aiwithkyle.com/join

Kyle’s AI Canon — Curated list of AI learning essentials: https://aiwithkyle.com/ai-canon

Subscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQg

This video is from the ‘AI with Kyle Live Stream’

First aired: 12 February 2026

Summary: Today's show was a bit different—I spent most of it reading through Matt Schumer's essay that's absolutely blown up on Twitter (71 million views, which is bonkers). The piece is called "Something Big is Happening" and it's basically Matt finally dropping the polite dinner party version of what's coming with AI and giving people the honest truth. He draws this parallel to COVID, how in February 2020 most of us weren't paying attention until suddenly the entire world changed in three weeks. His argument is we're in that same "this seems overblown" phase right now, except this is going to be much bigger than COVID. I've had this exact same feeling for ages—whenever I'm at parties and people say "AI can never do MY job," I just bite my tongue because polite company doesn't appreciate someone going hard on how their entire profession is about to be upended.

The essay covers everything from where we actually are with AI capabilities (spoiler: way further than most people realize) to what you should be doing about it right now. Matt talks about how a remarkably small number of people—maybe a few hundred researchers at OpenAI, Anthropic, and Google—are shaping the entire future, and how most companies are still using AI like it's a slightly better search engine. His main point is that there's a brief window right now where being the person who actually knows how to use these tools properly makes you the most valuable person in the room. I completely agree with his take that if your company is banning or pushing back on AI, it's done—might not feel like it yet, but in a couple years they'll be crushed. The whole thing is worth reading, so I'm putting it in tomorrow's newsletter.

Watch full edited live stream: https://youtu.be/VJwDbnga3J8

—— Time Stamps ——

0:00 Something Big is Happening — Matt Schumer's AI Warning

—— More Useful Resources ——

Find everything else at https://aiwithkyle.com/

Free OpenClaw set-up-guide: https://aiwithkyle.com/openclaw

Get AI-Ready with Kyle’s 5-Day Challenge: https://aiwithkyle.com/join

Kyle’s AI Canon — Curated list of AI learning essentials: https://aiwithkyle.com/ai-canon

Subscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQg

This video is from the ‘AI with Kyle Live Stream’

First aired: 11 February 2026

Summary: Right, so today's a bit different—no major AI news to report, just a flowery resignation letter from Anthropic that's more gossip than anything substantial. Instead, I'm diving into OpenClaw (or Claude Bot, whatever you want to call it) because whilst it's been out for a few weeks and absolutely exploded on GitHub—185,000 stars and climbing past React, closing in on Linux of all things—there's still loads of confusion about what you can actually use it for. Should you be using it? Are you falling behind if you're not? Spoiler: my mum asked me to set it up for her and I told her not to bother, so no, you're not obligated.

What I'm covering today is practical use cases rather than installation steps, because the real question isn't "how do I install it" but "what can I do with it that I couldn't do before." The key difference is that OpenClaw is always on—it's proactive rather than reactive, so you can set it up like an employee that autonomously handles tasks whilst you get on with other things. I'll walk through examples from the showcase and various meetups happening in San Francisco and Vienna, plus we'll touch on running local models if that's your thing, though fair warning: running something like Gemini 2.5 properly requires about a quarter million quid in GPUs, which is a bit steep for most of us.

Watch full edited live stream: https://youtu.be/Kyvh5G_BUxo

—— Time Stamps ——

00:00 Introduction: What is Clawd Bot & Should You Use It?

—— More Useful Resources ——

Find everything else at https://aiwithkyle.com/

Free Business Finder Quiz (used in video): https://aiwithkyle.com/tools/ai-business-quiz

Get AI-Ready with Kyle’s 5-Day Challenge: https://aiwithkyle.com/join

Kyle’s AI Canon — Curated list of AI learning essentials: https://aiwithkyle.com/ai-canon

Subscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQg

This video is from the ‘AI with Kyle Live Stream’

First aired: 10 February 2026

Summary: OpenAI's rolling out ads in ChatGPT and they're taking quite the interesting PR approach—essentially telling everyone that actually, you're going to love these ads. It's a bold strategy, I'll give them that. They're starting with free and Plus users in the US, promising the ads won't influence responses and will be clearly labeled. The spin is that this lets them expand free access to AI for everyone, which is fair enough given they need revenue. But you can tell from the comments that nobody's particularly thrilled about it. Still, I think it's reasonable—they've got to pay the bills somehow.

On a completely different note, I got properly annoyed with the UK government's AI training website. They've spent £4 million of taxpayer money on this thing through PwC, and it's genuinely useless. The "pathways" feature just shows you a filtered list of courses—it doesn't actually create a pathway at all. What's worse, they're missing obvious resources like NHS and civil service training materials that already exist. So I've gone ahead and started building a proper version myself that actually helps people navigate AI training, and I'll release it for free once it's done. Might replicate it for other countries too, because clearly someone needs to do this properly.

Watch full edited live stream: https://youtu.be/qRXXbU21fXk

—— Time Stamps ——

00:00 OpenAI Rolls Out Ads in ChatGPT

—— More Useful Resources ——

Find everything else at https://aiwithkyle.com/

Free Business Finder Quiz (used in video): https://aiwithkyle.com/tools/ai-business-quiz

Get AI-Ready with Kyle’s 5-Day Challenge: https://aiwithkyle.com/join

Kyle’s AI Canon — Curated list of AI learning essentials: https://aiwithkyle.com/ai-canon

Subscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQg

This video is from the ‘AI with Kyle Live Stream’

First aired: 9 February 2026

Summary: I spent today talking about something that's been bugging me—people asking permission to build things when they should just be shipping. Kit just tweeted about it, and I get the same thing constantly: DMs from people with "great ideas" who want to chat before they've built anything. With all due respect, I don't care about your idea. Come back with a link. The technical barriers are gone—vibe coding exists, you don't need a developer or a hundred grand to hire one. If you can't be bothered to prototype something using the tools available right now, that tells me you're not going to put in the work anyway. I'm literally rebuilding that embarrassing £4.1 million government AI Skills Hub myself just to prove the point.

The window is closing, and that's what people need to understand. Right now, you're ahead of the curve if you're watching this—you know AI exists, you understand what it can do, and you have maybe a year or two before everyone else catches up. I walked through a proper framework for finding your market, working out what business to build (hint: start with the business problem, not the AI), and why speed matters more than perfection. Peter Levels said it best—do you really have time to wait a few more years before starting? If we hit AGI, it's going to be exponentially harder to carve out a space for yourself. Build now, get distribution now, get customers now. Stop asking for permission.

Watch full edited live stream: https://youtu.be/xVGVQ4iQrHA

—— Time Stamps ——

0:00 Stop Asking Permission — Just Build Something

—— More Useful Resources ——

Find everything else at https://aiwithkyle.com/

Get AI-Ready with Kyle’s 5-Day Challenge: https://aiwithkyle.com/join

Kyle’s AI Canon — Curated list of AI learning essentials: https://aiwithkyle.com/ai-canon

Subscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQg

This video is from the ‘AI with Kyle Live Stream’

First aired: 6 February 2026

Summary: Honestly, I'm feeling a bit overwhelmed this morning, and that's saying something considering I wade through AI news every single day. Yesterday's release of Opus 4.6 has properly knocked me sideways—not just because the benchmarks are stellar, but because for the first time I'm viscerally feeling like AGI might actually be around the corner. I threw some heavy problems at it and watched it work, and suddenly all those CEO promises about artificial general intelligence don't feel like hype anymore. It's that "three sleepless nights" moment Ethan Mollick wrote about in Co-Intelligence, where you have to genuinely reckon with what this means for your work, your kids, and frankly, the entire human race.

The timing is wild because we've also got OpenAI dropping GPT 5.3 and Codex 5.3 in the same window, so the competition is absolutely cooking right now. Meanwhile, the stock market took a proper header yesterday as investors started processing what rapid AI advancement actually means for traditional business. I spent six hours yesterday testing these new models side-by-side, and Codex genuinely beat Claude Code on a website migration task, which I wasn't expecting. The step change from Wednesday to Thursday has been stark enough that even someone neck-deep in this stuff daily is having to take a breath and process it all.

Watch full edited live stream: https://youtu.be/l68NSRW3cRs

—— Time Stamps ——

0:00 Opus 4.6 Just Dropped — Is This AGI?

—— More Useful Resources ——

Find everything else at https://aiwithkyle.com/

Get AI-Ready with Kyle’s 5-Day Challenge: https://aiwithkyle.com/join

Kyle’s AI Canon — Curated list of AI learning essentials: https://aiwithkyle.com/ai-canon

Subscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQg

This video is from the ‘AI with Kyle Live Stream’

First aired: 5 February 2026

Summary: The AI drama is heating up. Anthropic just dropped an attack ad targeting ChatGPT's decision to include adverts in their free tier, which is a bold move considering Claude's relatively tiny user base. Sam Altman's response was perfectly brutal—basically saying "we're providing free AI to the world and using ads to fund it, meanwhile hardly anyone uses your product." Fair point, honestly. It's fun watching these tech giants take shots at each other, and ultimately it means better models and more choice for us.

Meanwhile, the UK government managed to spend £4.1 million (that's $5.5 million) on an "AI Skills Hub" that's essentially just a poorly organised catalogue of free courses you could've found on Google. The site's a mess—verification emails land in spam, there's no pricing information for paid courses, and I'm genuinely annoyed enough that I'm rebuilding it myself to show what £4 million should actually get you. On a brighter note, Google quietly revealed they've got 750 million users on Gemini, which means they're catching up to ChatGPT fast whilst everyone else is busy firing shots at each other.

Watch full edited live stream: https://youtu.be/JNqAvlt5tHk

—— Time Stamps ——

0:00 UK Government Wasted £4 Million on This AI Website

—— More Useful Resources ——

Find everything else at https://aiwithkyle.com/

Get AI-Ready with Kyle’s 5-Day Challenge: https://aiwithkyle.com/join

Kyle’s AI Canon — Curated list of AI learning essentials: https://aiwithkyle.com/ai-canon

Subscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQg

This video is from the ‘AI with Kyle Live Stream’

First aired: 4 February 2026

Summary: I spent the morning walking through OpenAI's new Codex Mac app, which is essentially their AI coding tool wrapped in a chat interface rather than a traditional development environment. The big news is it's free for the next two months, even on free ChatGPT accounts, which makes it properly accessible for people who've never touched code before. I wanted to show the complete setup process because whilst tools like Lovable are still easier, they cost £25 a month—and Codex being free removes that barrier entirely.

The tutorial covers the awkward bits that trip people up: setting up GitHub (using the desktop app, not the terminal—much simpler), getting your project deployed so you can actually see what you're building, and the general workflow of chatting with the AI to build something functional. I built a pattern-tracking app that analyses your daily habits over time, nothing fancy but it demonstrates the process. It's not the most technical guide out there, but it's designed to get beginners past those initial stumbling blocks so they can start making things.

Watch full edited live stream: https://youtu.be/H9HoL70WdPQ

—— Time Stamps ——

00:00 Introduction: What is OpenAI Codex?

—— More Useful Resources ——

Find everything else at https://aiwithkyle.com/

Get AI-Ready with Kyle’s 5-Day Challenge: https://aiwithkyle.com/join

Kyle’s AI Canon — Curated list of AI learning essentials: https://aiwithkyle.com/ai-canon

Subscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQg

This video is from the ‘AI with Kyle Live Stream’

First aired: 3 February 2026

Summary: OpenAI dropped their Codex desktop app for Mac yesterday, and the big news is that it's now available for free users—not just the paid crowd. They've also doubled the rate limits for a couple of months, which is pretty generous. I decided to take it for a proper test drive by building an RSS reader live on stream, inspired by Andrej Karpathy's recent tweet about getting back to RSS feeds and away from the endless slop on social media. The whole thing took less than an hour, and honestly, it worked pretty well. You can just install the app and start building without needing to mess around with GitHub straight away, which makes it far more accessible than something like Claude Code.

The verdict? It's solid. Not quite as polished as Claude Code in my view—it's slower, there's no preview panel, and you'll still need to be comfortable with the terminal for some bits. But for free access to AI coding, it's pretty remarkable. If you've been curious about vibe coding but didn't want to shell out £100-200 a month for Claude Code, this is your moment. The app has automations and skills (though skills are a bit broken at the moment), and you can run multiple projects in parallel. I'm going to keep building out my RSS reader in Codex as an experiment, mainly because it's free and I want to see how far it can go.

Watch full edited live stream: https://youtu.be/n60A8BlDzMc

—— Time Stamps ——

00:00 Intro: OpenAI Codex Desktop App Released for macOS

—— More Useful Resources ——

Find everything else at https://aiwithkyle.com/