In 2026, I expected to finally put my Claude Code subscription to good use by vibe-coding a text-based RPG I've always wanted to create. Instead, half the year has gone by, and Claude has helped me make something far more personal — my entire wedding plan. With a million different things to plan, it's impossible for an entire team of planners to do it right, let alone a sole AI agent, so I tried asking ChatGPT for help, too.

I didn’t think I would abandon NotebookLM. Ever since its introduction, it felt like magic. Like many power users, I fed it my PDFs, notes, and research documents, and let its neat summaries and famous Audio Overviews do the heavy lifting.

Anthropic has now built persistent memory into Claude Code. It works in two ways. First, there's Auto Memory, where Claude automatically saves useful project context, patterns, and preferences. Second, there's the CLAUDE.md file, which stores project instructions and context that get loaded into future sessions.

Google's AI Overviews are running into an interesting problem right now. Earlier on Friday, if you searched for the term "disregard," the AI Overview section would include a response like what you'd see from a more traditional AI chatbot instead of the typical AI summary, as spotted on X. As you can see in the image at the top of this story, I got an AI Overview response that said, "Got it. If you need anything else or have a new question later, just let me know!"As of Friday afternoon, however, Google isn't showing an AI Overview for the term "disregard" at all - instead, it shows a list of news stories about the issue first. Google hasn' …Read the full story at The Verge.

I've been a home theater fanatic for decades, with various setups in that time. I'm also a magnet for weird tech issues, including AC power quality issues in several places I've lived. More so in the U.S., where U.K. building standards mean every socket is properly grounded, but it could happen anywhere.

If you’ve worked on DIY computing projects in the late 2010s, you’ve definitely heard of the Raspberry Pi, if not own a few single-board computers belonging to this family. After all, their tiny form-factor, affordable price tags, and solid compatibility with popular Linux distros (and packages) made them the perfect tinkering companions. But that’s all in the past now.

For the longest time, the conversation around local AI models revolved around quality. They were either too slow, too dumb, too small, or too incapable to match what the titans over at OpenAI, Anthropic, and Google are doing with ChatGPT, Claude, and Gemini, respectively. That gap, however, is shrinking a lot faster than most people realize, even though it does exist in some areas. For the most part, though, modern local models have become genuinely impressive, and are capable of writing, summarizing, coding, and reasoning on capable hardware, of course.

If I remember correctly, I've had USB ports on my monitors since 2017, but I never thought of actually using them. The ports on the motherboard and case always proved to be enough, leaving no reason for me to consider the onboard USB ports on the monitor. When I switched to wireless peripherals, however, my monitor's USB hub became a lifesaver. It solved the annoying problem of wireless interference between all the dongles by allowing me to create enough distance between them. Even if you don't have wired peripherals, using the monitor's USB hub provides a more convenient and clutter-free way to connect them. Your monitor's USB ports are meant for convenience instead of peak performance, so stop ignoring them.

Between their dirt-cheap prices, terrific support for sensors, and minimal OS overhead, microcontrollers are a rock-solid choice for automation, circuitry, robotics, and other budget-friendly DIY projects where you might need real-time control over all the peripherals. While different MCU families have their own perks, the ESP32 lineup reigns supreme when it comes to sheer performance and connectivity options.

Twelve South’s AirFly Pro 2 let you connect wireless headphones to in-flight entertainment systems as well as the Nintendo Switch. | Image: Twelve South With Memorial day weekend kicking off the travel season, we’re seeing a lot of deals pop up on travel gadgets, from portable power banks to noise-canceling headphones. One of the best right now is Twelve South’s AirFly Pro 2 Bluetooth adapter, which lets you use your wireless headphones with in-flight entertainment systems so you can enjoy your flight a little more. It’s currently down to $49.99 ($10 off) at Amazon and directly from Twelve South, which is one of its best prices to date.Twelve South AirFly Pro 2Where to Buy: $59.99 $49.99 at Amazon $59.99 $49.99 at Twelve SouthThe Bluetooth transmitter lets you ditch the airline’s wired earbuds in favor of your own Bluetooth headphones or earbuds, which makes for a much better in-flight listening experience. All you need to do is plug the AirFly Pro 2 into the headphone jack on a seatback entertainment system, pair your headphones, and you’re set. It also supports two pairs of headphones at once, so you can watch movies or listen to podcasts with a travel companion. As Twelve South’s premium AirFly model, the Pro 2 also adds a few welcome improvements that make it even easier to use. That includes the cheaper AirFly SE’s dedicated onboard volume controls as well as an upgraded processor, which enables faster pairing and improved sound quality with less background noise. Its battery should also last up to 25 hours on a single charge, which should comfortably last you through even the longest flights.The AirFly Pro 2 remains useful long after you land, too. As it works with any standard audio jack, you can also use it to connect your wireless headphones to devices like a Nintendo Switch , older car stereos, and even compatible gym equipment like treadmills.

The AMS has always been one of Bambu Lab’s best ideas, but also one of its messiest workflows. It can swap materials, keep spools ready, and turn a single printer into a much more flexible machine. Yet the software side has never quite felt as polished as the hardware promised. For a system built around loading multiple spools at once, keeping track of those spools has often been weirdly manual.

Most of the small reasoning models that have shipped in the past year are variations on a theme. A familiar transformer backbone, a Mixture-of-Experts wrapper, grouped-query attention or something like Gated DeltaNet in Qwen's case for a smaller KV cache, and a heavy reinforcement learning stage at the end. Performance improves year on year, but the architecture of what's actually running is similar to the shape it was when DeepSeek R1 arrived.