The other day I saw my friend Larry Hyrb, AKA Xbox’s Major Nelson showing off his Native Union POP Phone and I knew I had to get one too. It’s a telephone handset that looks like the same one my family had hanging on the kitchen wall when I was a kid. While it might look like something out of a time capsule, it’s actually built for the modern era with a USB-C connector that lets it plug into today’s phones, tablets, and computers. You can find the phone on Amazon (compensated affiliate link).
It definitely lacks that heavy, metal feel the old ones did — this one’s all plastic, lighter, and sealed up, so there’s no unscrewing the receiver to see what’s inside. The cable is coiled like the originals but permanently attached, and there’s a single button for hanging up calls. It’s not quite as satisfying as slamming a handset onto a cradle, but if you need the full experience, you can always tap the button and smack it on the desk for good measure — just gently.
To really put it to the test, I called up Larry on a video call which you can see in the video. On my end, the audio setup didn’t cooperate perfectly, but Larry’s side of the call gave a clear sense of what the phone sounds like — surprisingly better than expected, with decent noise isolation.
We both laughed about the muscle memory that came right back — the way you naturally cradle it between your ear and shoulder or reach for a coiled cord that could stretch halfway across the house. Larry mentioned using his Pop Phone in meetings just to throw people off. Older coworkers immediately recognize it, while younger ones have no idea what they’re looking at. It’s one of those small reminders of how quickly technology moves on from something that used to be a fixture in every home.
The Pop Phone is a novelty for sure. But it’s a gag gift that’s actually pretty functional and works with just about any device that has a USB-C port.
After last week’s video about how smart TVs spy on users, I wanted to take a deeper look at the legalities around allowing TV manufacturers to spy on everything we watch – including what’s connected to our TVs via the HDMI port.
As a recap, most televisions don’t just track what apps you use—they can identify what’s on the screen or what’s coming through the speakers, then send that data off to advertisers and data brokers. It’s all done through automatic content recognition, or ACR, and it’s completely legal because users consent to it, often without understanding they have.
When I factory-reset my Roku TV, the setup process gave me two options in regards to ACR: “Agree” or “Manage Preferences.” There was no simple “Yes” or “No.” Most people, eager to get started, are going to hit “Agree.”
If you do click through to “Manage Preferences,” you can then opt out, and Roku will still let you use its smart features. That’s more than I can say for my LG TV, which shut down all its smart functions when I declined a new privacy policy after a firmware update. I could still use connected devices, but the built-in apps were locked out until I accepted the new terms. Roku’s approach at least lets you continue using the interface, but I doubt many users go through the trouble to opt out. A real opt-in should offer a clear yes-or-no choice, not bury “no” under layers of menus.
Roku’s privacy policy itself is over a hundred pages long printed out, and scrolling through it takes several minutes. Buried in that text are all the details about how the company collects and sells data. The numbers make it clear why this is so central to their business—Roku’s recent quarterly report showed more than a billion dollars in gross profit from its platform, compared to only about $146 million from hardware. The TVs are just the delivery mechanism; you and your data are the product.
Apple has taken the opposite approach by asking users directly whether they want to be tracked across apps. The first choice shown is “Ask App Not to Track,” followed by “Allow.” When Apple rolled this out, 96 percent of U.S. users opted out, and even now most people still refuse tracking when given a clear choice. Reports from analytics firms put the current opt-in rate somewhere between 15 and 30 percent.
Looking ahead, I’m concerned about where this technology might go as AI becomes more powerful. Right now, companies say they’re only sending “fingerprints” of screen images, not the images themselves, but even small local models that can run on smartphones analyze photos in surprising detail. It’s easy to imagine a manufacturer deciding that full-image uploads could make targeting more precise and profitable.
Many viewers told me the simple answer is to keep TVs offline. I agree—that’s the easiest fix. Unplug the Ethernet cable, disable Wi-Fi, and use an external device like an Apple TV or a computer if you want streaming apps. But most consumers don’t do that. When I stopped by Best Buy recently, the salesperson said people mainly care whether their new TV supports the apps they use most. They’re connecting their sets because they want convenience, not because they’ve read a privacy policy.
If regulations ever catch up, maybe they’ll require true opt-in choices instead of manipulative prompts. Until then, the safest move is still to disconnect your television from the internet and think carefully about what you’re agreeing to.
For a good resource on taking back control, my friend Veronica over at Veronica Explained has a video on cutting these services out entirely and running everything with open-source tools. She’s got some solid ideas for handling your own streaming setup without giving away your data.
I took a train into New York City on Wednesday to check out Pepcom’s annual holiday preview show. It’s always a little strange to see Christmas decorations everywhere before Halloween, but this event serves as an early look at the gadgets companies are hoping will land on holiday wish lists. I walked the floor to see what stood out.
Withings was there showing off the U-Scan, a health device designed to fit inside a toilet bowl. It analyzes urine to track things like nutrition and kidney health. The system uses cartridges that last about three months and can identify multiple users automatically. There’s also a cleaning and charging station to keep it maintained. It’s a bit unconventional, but it’s another example of how personal health tracking keeps expanding into new spaces.
Remarkable had a smaller version of its popular e-ink tablet called the PaperPro Move (compensated affiliate link). It keeps the same minimalist design but shrinks things down to a more portable size that fits in the hand. It’s designed for people who prefer a distraction-free writing experience without the bulk of a larger tablet.
A company called Neurable caught my attention with a pair of headphones that double as a brainwave reader. The silver pads on the band are electrodes that can analyze brain activity to estimate your “brain age” and track changes over time. They’re also regular headphones, but the focus here is on cognitive health rather than just audio quality.
Xperi—better known for TiVo—demonstrated its television operating system that’s beginning to appear on Sharp Aquos TVs (compensated affiliate link) in the U.S. The interface organizes content recommendations across streaming apps, including Plex, and aims to make finding something to watch a little easier.
Kensington had a few new items, including the Expert Mouse TB 800 EQ trackball. It has adjustable polling rates, multiple buttons, and a sturdy metal design with a satisfying scroll dial. They also showed off a Thunderbolt 5 dock loaded with ports, an M.2 slot for storage, and a compact flash card reader. It’s hefty, but that makes it stay put on the desk. These aren’t shipping just yet but will be available soon at Amazon (compensated affiliate link).
Ambient was demonstrating a bedside clock called the Dreamie that encourages better sleep habits without relying on a phone. It uses light and sound cues to signal bedtime and can play white noise, podcasts, or ambient sounds through the night. A motion sensor tracks how often you get up and records environmental conditions like humidity and temperature. It’s designed to promote healthy sleep without constant screen time.
HP had a lightweight 14-inch portable display that covers the full sRGB color range. It’s bright for a bus-powered monitor and includes a magnetic cover that doubles as a stand. They also had a new version of their HyperX SoloCast 2 microphone (compensated affiliate link), a simple USB mic that’s inexpensive and aimed at beginners who want better sound for podcasting or streaming.
Another HP product on display was a League of Legends–themed Omen laptop, complete with game-inspired design details and high-end specs. It’s targeted at fans of the game who want hardware that matches their interests.
Yubico didn’t have anything brand new, but they noted that their security keys are now available in Best Buy stores (compensated affiliate link), which makes it easier for people to pick one up locally.
Nvidia was at the show too, but there was no sign of a new Shield device—just their GPUs this time.
SanDisk, however, had a few storage products on display, including an ultra-small USB-C drive that fits flush with a laptop port and can transfer data at around 400 megabytes per second. They also had Crayola-branded USB drives (compensated affiliate link) shaped like crayons, with a few colors exclusive to specific retailers like Walmart.
I also came across Napster at the show, though this version of the brand has nothing to do with music. The company is introducing a new AI service built around access to specialized AI experts. You can ask questions and get responses based on the expertise of these AI personalities, and they’re designed to perform certain tasks as well. Napster is also showing a hardware accessory that clips onto the top of a laptop and projects a 3D version of your AI assistant, adding a visual layer to the interaction. But it’s Napster in name only.
That wrapped up my walk through Pepcom’s holiday tech showcase. I’ll be back at CES in January to see what comes next. This event always feels like the calm before the storm—an early glimpse at what we’ll soon see everywhere once the holidays hit full swing.
I’ve been running my Plex server on Unraid for a while, and it’s become one of my favorite setups for hosting my media library. Unraid makes it simple to manage because it runs applications inside Docker containers, which keeps things clean, organized, and portable. One other neat feature of Unraid is how easy it is to migrate to new hardware: you can just move your drives and boot device over to the new system, power it up, and everything comes right back up. Docker containers are portable too, so even if you migrate off Unraid later, your Plex setup can move right along with you.
To get started, I went into Unraid’s app section and searched for Plex. There are a few different versions available, but I used the official container maintained by the Plex team. Installing it was just a matter of filling out a few fields in the template and clicking “Start.” For the network setting, I left it on “host” so Plex could be easily discovered by devices on my local network. If you use bridge mode, you’ll need to set up port mappings manually, but for most home setups, host mode is fine.
Before launching the container, I created a file share on my Unraid server called “Plex Media.” Inside it, I made folders for data, transcode, movies, and TV. The data folder stores Plex’s metadata, transcode is where temporary files go during playback, and the media folders hold the content itself. When setting up the container, I mapped those folders accordingly. One extra step that’s important if you want hardware transcoding is to add a device mapping for /dev/dri, which points Plex to the hardware encoder on your CPU. Without that, Plex won’t be able to use Intel’s Quick Sync for transcoding.
You’ll also need to claim your Plex server with a code from plex.tv/claim. The code only lasts a few minutes, so it’s best to grab it right before launching the container. Once the container starts up, Plex downloads and configures itself automatically, and then you can connect to it from a browser using your Unraid server’s IP address followed by port 32400.
From there, setting up the libraries is straightforward. I added one for movies and another for TV shows, each pointing to the appropriate folder inside the media directory. Plex scanned the files and quickly pulled in the metadata for Back to the Future and Stargate SG-1, which I’d placed there as test files.
To make sure hardware transcoding was working, I played Back to the Future on my phone and forced it to transcode from 4K HDR down to 1080p. Watching the Unraid dashboard, I could see that CPU usage stayed low while the Intel N150 processor handled the conversion efficiently through its onboard video encoders.
Keeping Plex updated on Unraid is also simple. You can do it through the Apps tab—Unraid will notify you through the “Action Center” when an update is available—or you can update directly from the Docker section. It’s just a couple of clicks either way.
Running Plex on Unraid combines the flexibility of Docker with a simple interface that’s friendly to newcomers but powerful enough for tinkerers. If you decide to move your setup later, you can take the entire Unraid install or just the Plex container to another machine without much effort. And once you’re comfortable, Unraid opens up a whole world of self-hosting possibilities beyond Plex, with tons of other apps available that install just as easily.
Disclosure: This was a paid sponsorship for Plex. They did not review or approve this content prior to upload and all opinions are my own.
When I bought my LG OLED TV about eight years ago, I never imagined it would one day be spying on everything I watched. Like most people, I was aware that smart TVs track viewing habits for marketing purposes, but what I didn’t realize until recently is just how deep that surveillance goes. These devices actually capture images and audio from anything connected to the TV, whether it’s a game console, a streaming box, or even a home movie streamed from your phone. That information gets packaged up and sent to data brokers or used to target ads across the web.
This kind of tracking happens through something called Automatic Content Recognition, or ACR. It works by sampling what’s on the screen, matching it against a database, and then building a profile around what your household watches. This data is also used to help marketers see how many viewers actually see their ads.
When I went through the privacy settings on my LG set after a firmware update, I discovered the TV was monitoring all HDMI inputs, not just built-in apps. And when I tried to opt out, the TV refused to let me use any of its “smart” features unless I agreed to those terms.
Other manufacturers handle it differently, though not necessarily better. Samsung buries its ACR disclosure deep in its privacy statements, and while there’s an option to disable “SyncPlus and Interactive Functions,” it’s not clear how complete that shut-off really is.
Amazon’s Fire TV–powered televisions create digital fingerprints from the shows and ads you watch, saying the goal is to verify ad impressions and “reduce repetition,” but that still means every pixel and sound might be analyzed.
Roku is the most open about its practices – and even brags about winning an Emmy for their TV spying technology – mostly because it uses that transparency to sell advertisers on the value of its data. The company even boasts about its ability to track what games are being played on connected consoles and for how long people play them.
Google TV is the biggest mystery of the bunch. There’s little public information about whether Google itself runs ACR or leaves it to each manufacturer. HiSense, for instance, admits to collecting both audio and video data through its Google TV sets. I couldn’t find any comparable details from Sony (a larger maker of Google TV sets), which suggests the fine print may only appear on the TVs themselves, hidden behind those long on-screen agreements few people read before clicking “accept.”
For anyone worried about this kind of data collection, the best defense is to treat your TV as just a display. Disconnect it from the internet and use a separate streaming box instead. I use an Apple TV for that reason—it isn’t perfect, but it’s far less aggressive about data sharing than the others. Consumer Reports maintains a useful guide explaining how to disable tracking features across most major brands, which I’d recommend checking out.
After reading through my LG’s privacy policy line by line, I was startled to realize how much of my personal life could be analyzed simply because it passes through an HDMI cable or streamed to it over my local network. The notion of “the privacy of your own home” is quickly becoming eroded by our “smart” technologies.
I picked up the second-generation ADTH ATSC 3.0 NextGen TV tuner on Amazon the other day (compensated affiliate link) to see how it performs. If you’re not familiar with it, the device lets you receive both traditional ATSC 1.0 and new NextGen TV broadcasts, including those that use DRM encryption. It can record encrypted broadcasts to an SD card, though playback only works on the box itself. ADTH also claims that a “gateway” firmware update will allow more flexibility later.
What caught my attention was the box’s promise of “unconnected DRM,” suggesting it wouldn’t need an internet connection to tune in live television. The first-generation model required one, which made no sense for free over-the-air TV. So I decided to test that claim directly—no Wi-Fi, no Ethernet, just power and an antenna.
Out of the box, the tuner booted up and I was able to bypass the WiFi setup. I manually set the time and region since there was no internet connection to do it automatically. That step alone might frustrate less technical users. The channel scan found about 56 stations, and I was able to tune both encrypted and unencrypted channels without connecting online. My local encrypted CBS and NBC affiliates came through fine, confirming that offline decryption now works.
Unfortunately, HDR output was dim on my non-HDR display. There’s no option to disable HDR in the settings—just “always” or “adaptive”—and color-space adjustments made no difference. That means the firmware still forces HDR on displays that can’t handle it.
Recording worked in a limited way. I could start a recording to the SD card and even switch channels on the same broadcast frequency, but the box only supports a single tuner. Once a recording finished, I couldn’t find where it went. The “library” section that should have contained my recordings was missing until I connected to the internet and performed several beta firmware updates. ADTH is advertising DVR functionality and offline use simultaneously, but in practice, the DVR only becomes usable after downloading updates and enabling a beta mode.
After a few updates, the library finally appeared and playback worked, including for encrypted channels. It did play back those encrypted recordings when I disconnected it from the Internet. However, the software remains buggy—the interface froze at one point, requiring a power-cycle. The tuner also runs an old version of Android 11 with a 2021 security patch, making it dangerously out of date.
So, while the device technically works without the internet for basic live TV—including DRM channels—most of its useful features require network access. The HDR bug, outdated software, and rough interface make it feel unfinished. It still costs $100, far more than the $20 tuners available for ATSC 1.0, largely because of the licensing fees and development costs tied to encrypted broadcasts.
This is one of the few options that can handle NextGen TV’s DRM, but it’s hard to imagine a casual viewer getting this up and running smoothly. I’ll keep testing it as updates roll out, but this second-generation box feels like another rough draft in the long transition to ATSC 3.0.
In my latest video, I took a look at the new 2.0 firmware update for the 8BitDo Pro 3 and Ultimate 2 Bluetooth controllers (compensated affiliate links), both of which can now wake up the Nintendo Switch 2 with just a shake. This feature had been missing from earlier versions, so I wanted to go through what’s required to make it work. It does require a specific procedure, so a general pairing with the console won’t wake it up.
First, it’s important to note that only these two controllers — the Ultimate 2 Bluetooth Edition and the Pro 3 — currently support this wake function. Other 8BitDo models will connect to the Switch 2, but they can’t wake it up yet.
To get started, you’ll need to update your controller’s firmware to version 2.0 using the 8BitDo Ultimate software on your computer. Once connected via USB, the software gives you the option to update the firmware, which enables the new “shake to wake” feature. After the update is complete, the controller will reboot automatically.
Next comes the special procedure: First, make sure your Nintendo Switch 2 console is in sleep mode. Next, while the Switch 2 is asleep, put the 8bitdo controller into pairing mode by pressing the button on top until it starts blinking. Then, remove a Joy-Con from your Switch and power the console on using the Joycon’s Home button. From the main menu, go to the controller settings and select “Change Grip/Order.” At this point, press the left and right buttons on the 8BitDo controller until it pairs successfully.
After pairing, you can reattach your Joy-Con, put the Switch to sleep, and then try waking it up by turning on the 8BitDo controller and giving it a quick shake. Be sure the controller is set to Bluetooth mode rather than the 2.4GHz wireless dongle mode. On the Ultimate 2, make sure the mode switch is in the bluetooth position, and on the Pro 3, confirm that the “S” mode is selected.
Disclosure: 8BitDo and/or their distributor AKNES sent me these controllers free of charge. No other compensation was received, and they did not review or approve this content before it was uploaded.
My latest video review looks at the new Minix Elite ER936-AI Mini PC. It’s powered by AMD’s Ryzen AI 9 365 processor, part of the Strix line, which gives it ten cores and a surprisingly capable built-in GPU. My review unit came configured with 32 gigabytes of DDR5 memory running at 5600 MHz and a one-terabyte NVMe SSD. You can find it over at Amazon, or direct at Minix’s online store (compensated affiliate links).
I’ve reviewed a number of Minix devices before and have generally found them to be well built, and this one fits right in with that reputation. It’s designed to be easily serviceable, with a tool-less bottom panel that lifts off for access to the internals. Underneath, you’ll find an extra NVMe slot, the Wi-Fi card, and RAM slots.
The port selection is generous for a small system. There are two 40 gigabit USB 4.0 Type-C ports that can handle Thunderbolt and video output, plus multiple 10-gigabit-per-second USB-A ports, dual 2.5-gigabit Ethernet jacks, HDMI, DisplayPort, and a headphone/mic jack placed on the front. Power comes from a 120-watt adapter, and the system draws around 100 watts under heavy load but sips about 10 to 12 watts at idle. A small fan keeps things cool; it’s audible under stress but not particularly loud, and it does a good job of maintaining thermal consistency.
The system ships with Windows 11 Pro preinstalled and activated. General performance was solid across everyday tasks like web browsing and streaming, though I noticed a few dropped frames while playing 4K60 YouTube content. The machine scored 33 on Browserbench’s Speedometer test, putting it in line with other recent AMD and Intel processors. Its graphics capabilities stood out in particular. I tested DaVinci Resolve with 4K footage and saw smooth real-time playback of transitions and a noticeable speed advantage when rendering effects compared to other mini PCs I’ve tested.
I also experimented with local AI tools to see how it handled workloads beyond typical productivity. Using AMD’s GAIA and Lemonade applications, I ran a 20-billion-parameter open-source language model that utilized the GPU. It performed significantly faster than running similar models through Olama on the same hardware which only runs through the CPU. It’s not going to rival cloud-based AI tools, but the fact that it handled local LLM this well shows how far these compact systems have come. For anyone experimenting with automation tools like N8N or other LLM setups, this configuration offers a good entry point.
Gaming performance was better than expected. In Cyberpunk 2077 at 1080p low settings, frame rates hovered around 45 to 55 frames per second, dipping slightly in more complex scenes. No Man’s Sky performed even better, staying close to 60 FPS on the ground and climbing to around 80 in space. The GPU is the limiting factor, but there’s enough CPU headroom for an external GPU setup if you wanted more performance. On 3DMark’s Time Spy benchmark, the system scored 3,504, which puts it close to older desktop GPUs like the Nvidia GTX 1060 and 1650. The stress test showed stable cooling with a 99.3 percent consistency score.
Linux support was also solid. Ubuntu detected all of the system’s components, including Wi-Fi, Ethernet, and audio, without issue. With two NVMe slots, it’s easy to set up a dual-boot configuration for Windows and Linux. This flexibility makes it suitable for home server use or software testing environments where you want both operating systems available.
After spending some time with it, the Minix Elite ER936 feels like a well-balanced small form factor computer that can handle a mix of productivity, light gaming, and local AI tasks. The design is practical, it runs efficiently, and it delivers consistent performance under load.
Amazon’s new “Select” 4k streaming stick with the new Vega OS has not been well received – especially by enthusiasts. In my latest video, we take a look at what’s going with the FireTV and why Amazon is moving away from the Android player we’ve come to know and mostly love over the last decade.
When I started covering tech on YouTube more than a decade ago, one of the earliest products I reviewed was the original Amazon Fire TV. It was a time when streaming boxes were still new and fragmented. Roku was around, but like today it was very limited in capabilities, and Apple’s TV box didn’t yet have apps. Amazon’s entry in 2014 was a surprise — an Android-based device with an interface built for television. It even beat Google’s Nexus Player, the first official Android TV device, to market by a few months.
Back then, the Fire TV felt like a meaningful step forward. Amazon had invested in game development studios and the box had decent graphics performance for casual play. You could sideload Android apps, and it was fast at launching video, caching streams so they started almost instantly. The platform was flexible, and the company was building a product that appealed to both mainstream users and enthusiasts.
Fast forward eleven years, and Amazon’s latest Fire TV device, the 4K Select, runs something entirely different. The operating system, called Vega OS, has replaced Android under the hood, but Amazon isn’t marketing it openly. It’s not mentioned on the box or in promotional materials. What’s more, this new system limits what the device can do. Apps now need to be rewritten for Vega OS, and many haven’t made the jump yet. In some cases, Amazon is actually streaming apps from the cloud to make them run on the new hardware, a workaround that shows how much compatibility has changed.
This move appears to be a shift in priorities. Vega OS likely helps Amazon build cheaper hardware with lower overhead, targeting the low-end streaming stick segment rather than the higher-performance devices that used to appeal to enthusiasts. Developers can build in React Native, which is cross-platform, but that still means maintaining another version of their app specifically for Vega. Whether streaming app makers will see that as worth the effort remains to be seen.
The timing of this change may have something to do with where Amazon stands in the streaming device market. Data from Pixalate shows Roku leading with about 36 percent of U.S. market share, far ahead of Fire TV’s 14 percent. Roku focuses almost entirely on delivering video streaming with a simple interface. Consumers seem to prefer that over devices that try to do more. Fire TV’s more advanced features don’t appear to be helping it compete.
Roku’s financials tell a similar story. They’ve been selling hardware at little or no profit but making nearly a billion dollars a quarter in gross profit from their platform business — most of it advertising. These devices aren’t meant to be powerful computers anymore; they’re ad platforms with remotes attached. Amazon seems to be trying that model, prioritizing simplicity and scale over capability.
Google is reportedly rethinking its own TV strategy as well, possibly moving away from its current Google TV platform. For users who enjoyed the flexibility of older devices like the NVIDIA Shield (compensated affiliate link), there may not be many options left. The Shield still offers features like sideloading, local media playback, and advanced home theater support with Dolby Vision and lossless ATMOS, but it’s starting to look like an artifact of a different era.
I find it telling that Amazon, a company that once encouraged experimentation on its Fire TV line, is now quietly locking it down. For people who use these boxes just to stream Netflix or Prime Video, that may not matter. But for those who like to tinker — to run emulators, custom apps, or personal media servers — this marks the end of an era. The industry seems to be moving toward simpler, more disposable devices designed to serve ads and stream content, not extend functionality.
My advice? Buy as many NVIDIA Shield devices as you can while they’re still for sale.
I’m not someone who usually likes to stick things to the back of my phone, but about two years ago I came across the OhSnap Grip 4 (compensated affiliate link) and found myself not wanting to take it off. You can see my original review of it here.
It was a simple little accessory that functioned like a PopSocket—you could pop it out for a better grip or use it as a stand—and when you didn’t need it, it folded down super flat. What made it stand out to me was that it added MagSafe-style functionality to Android phones through its magnetic ring. It wasn’t something I expected to keep using, but it became part of my daily routine.
Now the company has released an updated version, the OhSnap 5 Grip (compensated affiliate link). It largely looks the same but it no longer uses the strong but removable adhesive of the prior model. OhSnap says they very carefully designed a stronger magnet system that hangs on tighter.
To use the new model on Android phones, the company includes a metal ring that gets stuck to the back of the phone. Once attached, the grip will snap neatly into place and retains the same features as before—the pop-out stand and the comfortable finger grip. But Android users will have to pry lose the adapter ring if they stop using the Snap 5 grip.
The design looks thin but feels sturdy, just like the last one, though it’s slightly thicker. When I tested how securely it attached to my caseless iPhone 17 Pro Max, it came off more easily than the adhesive model. A firm tug or some lateral pressure will very quickly snap it off. For someone like me who uses a phone without a case—my running joke is that AppleCare is my case—the old adhesive-backed version feels much more secure.
When I tried it on my iPad mini with a metal Magsafe adapter ring installed, the grip was much stronger—so much so that I couldn’t pull it off easily. That told me the performance really depends on the phone’s case and how close the case’s magnetic ring sits to the surface.
Like the earlier version, the Grip 5 is MagSafe compatible and allows for chargers and other MagSafe accessories to snap on top. While MagSafe devices snapped on as expected, the strength of the outer magnets is noticeable weaker than the Grip 4. For charging, the Grip 5 works best with MagSafe chargers but doesn’t pair well with standard Qi chargers that lack magnetic alignment.
After a few days of use, I’ve found that while the Grip 5 is not as good as its predecessor. The magnet is strong but not strong enough to serve as an effective substitute for the reusable adhesive, and its outer magnet does not latch onto MagSafe accessories as confidently as before.
I’ll therefore be sticking—literally and figuratively—with the old one.
Disclosure: I purchased a Grip 5 with my own funds, and the company separately sent me another one free of charge. No other compensation was received, and the company did not review or approve this content before it was uploaded. All opinions are my own.
Amazon’s latest Fire TV Stick 4K Select (compensated affiliate link) marks a shift in direction for the company’s streaming devices. This one runs on a completely new operating system called Vega OS, replacing the Android-based OS Fire TVs have used over the last decade. But at launch, it’s clear this is a minimally viable product, and early adopters will need to temper expectations.
As far as the product lineup goes, this device sits just above the 1080p stick and just below the full-featured Fire TV Stick 4K. For now, I’d say it’s worth paying a little extra for the latter if you want a smoother experience. The Select model still uses micro USB for power, has just one gigabyte of RAM, and supports Wi-Fi 5 instead of Wi-Fi 6. It also lacks support for Dolby Vision which is found on the other 4k sticks.
Once set up, the interface looks familiar if you’ve used a Fire TV before. Most of the big-name streaming apps are available, but not all of them are native to the device. Some, like the NFL Network, are cloud-hosted, meaning both the app and its content stream over the internet. It works, but performance can feel sluggish, even on a fast connection. Specialized apps for devices such as the Tablo or HDHomeRun aren’t available yet, and while Plex is present, it’s missing many of the features found on its Android-based counterparts.
App support overall feels limited. There are no games and no sideloading options. The experience is closer to a Roku, but with fewer apps to choose from. For basic streaming, though, Netflix, Disney+, and YouTube run well enough. The device supports basic HDR (but again not Dolby Vision), and Dolby Atmos audio works through compatible services.
Performance on YouTube was mostly solid with 4K HDR playback, though I had to manually force it into 4K60 mode. There were occasional frame drops, but nothing too disruptive.
One issue that stood out was the lack of proper 24p output for films and TV shows in all of the apps I tested. Even with the “match frame rate” option enabled, the device stayed locked at 60Hz, resulting in uneven motion that film buffs will notice.
Voice control works as expected, with Alexa handling searches and smart home commands reliably. The remote can also control basic TV and receiver functions, though support for other devices appears trimmed down compared to earlier models.
At this stage, the Fire TV Stick 4K Select and its new Vega OS feel like a work in progress. It handles core streaming tasks fine, but beyond that, the limitations add up. This isn’t the Fire TV experience power users have come to expect—it’s more locked down, less flexible, and missing key features found on cheaper models. Amazon’s long-term plan is clearly to transition its lineup to this new platform, but until app developers catch up, it’s going to feel like a step back.
For now, it works if all you need are the major streaming apps on an older TV. But as Amazon moves further away from Android, this early look suggests the future of Fire TV could be more controlled and less open than before.
The Thermal Master P3 is a thermal camera designed to plug directly into an iPhone or Android device through the phone’s USB-C port. While it’s a little on the pricey side it does offer some great features and has some useful macro features.
You can find one over at Amazon (compensated affiliate link). Once connected and the companion app is installed, the camera functions as a macro thermal imager—it’s clearly optimized for close-up inspections rather than broad outdoor scenes. It’s the kind of tool you’d use to check an electrical panel, a circuit board, or in my case, a Raspberry Pi, rather than searching for animals in the woods.
It can be attached facing either direction and even accommodates some thicker phone cases thanks to its extended connector, though not every case will fit comfortably. For older iPhones, there’s a lightning adapter included. When using the lightning adapter, the setup feels a bit unstable, so a short extension cable might be a better option.
There’s a focus ring built in, which helps bring small components into sharper view. The sensor itself is very small, with a native resolution of 256×192. The app offers an AI-enhanced mode that upscales to 512×384, but even then, fine detail is limited. The best results come when the camera is positioned close to the subject.
While testing, I filmed my Raspberry Pi booting up and watched the components warm in real time. I also pointed it at a stovetop to see how it handled extreme temperatures. The app includes built-in burn-in protection that reduces exposure to high heat sources, prompting me to back off a bit. Temperature ranges can be adjusted within the app, with one mode for lower temperatures (-4 °F to 302 °F) and another for higher ones (212 °F to 1,022 °F). You can’t combine the two ranges, but for most household or electronics projects, the preset bands are sufficient. A scale overlay can be toggled on to display minimum and maximum readings in real time.
I experimented with the camera around the house—tracking hot water lines, scanning my electrical panel, and exploring different color modes. The app provides several palette options such as white hot, black hot, and various false-color schemes. There’s also a split-view mode that overlays a standard camera image to help identify what you’re looking at when the thermal view alone doesn’t offer enough definition. Users can even draw shapes on-screen to isolate temperature readings from specific areas—useful for monitoring different components within the frame.
Although it costs less than dedicated professional thermal imagers, the device isn’t meant for scientific precision. The manual includes some reference values for emissivity correction, but calibration is limited, and readings shouldn’t be treated as exact. What it does well is provide quick visual confirmation of heat distribution. It’s compact, fits easily in a pocket, and draws power directly from the phone. For spotting hot spots on a breaker panel or verifying that a circuit board component isn’t overheating, it’s a handy, portable option that fills the gap between hobby gear and higher-end tools.
Disclosure: The camera was sent to the channel free of charge. No other compensation was received and no one reviewed or approved this content prior to uploading. All opinions are my own.
In my latest video review, I take a look at the My Arcade Atari GameStation Go (compensated affiliate link), a portable emulation console aimed squarely at those of us who grew up in the era of wood-paneled consoles and CRTs.
It’s a self-contained handheld that plays hundreds of mostly Atari classic games from the 80s and 90s and even lets you load more through an SD card. What caught my attention was the variety of control options—there’s a built-in paddle for games like Breakout, a trackball for Centipede and Missile Command, and a solid D-pad with shoulder buttons. It doesn’t have the high-end feel of a Steam Deck or Switch, but it manages to strike a decent balance between nostalgia and functionality.
Navigating the interface is straightforward. The menus are organized by system, with sections for Atari 2600, 5200, and arcade titles. I tried Breakout first, which uses the paddle controller, and I liked that the device’s “SmartGlow” lighting highlights which controls are active for each game. The paddle isn’t a free-spinning spinner, but it has enough resistance to make gameplay feel natural. The trackball, however, felt scratchy and inconsistent, and I found myself preferring the D-pad for precision control in Centipede. The D-pad itself is surprisingly good—firm, responsive, and better than my initial impression led me to believe.
The system includes about 200 licensed games, mostly from Atari’s catalog. Activision classics like Kaboom and Pitfall! are missing due to licensing, but loading them from an SD card is simple. Once I inserted a card, a hidden menu appeared, letting me boot ROMs from different systems, including the Sega Genesis, Super Nintendo, and even PlayStation. Performance varies—Sonic the Hedgehog lagged a bit, and Dreamcast games crawl. But PlayStation and Atari Lynx emulation worked better than expected. It’s essentially running an open-source emulator framework under the hood, so there’s potential for the community to expand its capabilities over time.
Battery life felt similar to a small tablet—around four to six hours, depending on what’s running. There’s HDMI output too, but connecting it reboots the system and reduces performance. The image quality looks fine at 1080p, but frame rates dip with some of the newer “Recharged” Atari titles installed on the console. Still, older Atari games like Bowling and Breakout play smoothly both on the built-in screen and an external display.
While this isn’t a machine for hardcore emulation fans, it’s an enjoyable, self-contained throwback for casual players and gift-givers who want a plug-and-play experience without tinkering. The stand-out features here are the paddle and trackball that allow for game controls that more closely match how these games were originally played.
The Raspberry Pi 500 Plus is a new take on the age-old keyboard-computer concept that combines a Raspberry Pi 5 with a mechanical keyboard and built-in NVMe storage.
In my latest video review, we take a look at this retro inspired offering from the Raspberry Pi foundation.
It’s selling for about $180 to $185 at official Pi retailers, which is roughly $100 more than the earlier Pi 400. Out of the box, it feels like the most polished of the Pi-in-a-keyboard systems I’ve tried, with better specs and a more comfortable design overall.
Inside, it has 16GB of RAM and a 256GB NVMe SSD that ships with Raspbian OS preinstalled, so it boots right up without any tinkering. You can, of course, use the microSD slot to load other operating systems.
Still, some of the frustrations from the Pi 5 carry over. The two micro-HDMI ports feel unnecessarily small given the space available on the back, and the power situation is still tricky. It requires a 5-volt, 5-amp USB-C power supply, which isn’t as common as you’d think. Most USB-C chargers top out at 3 amps in their 5 volt mode, and using one of those will trigger low-power warnings when you plug in peripherals and possibly lead to performance issues under load. Even Apple’s chargers don’t work properly with it, so it’s worth getting the official supply. I picked up one of these low cost ones at Amazon (compensated affiliate link).
Ports include two USB 3.0, one USB 2.0, Gigabit Ethernet, and the familiar GPIO header under a rubber cover. AC Wi-Fi and Bluetooth are built in, and dual 4K60 HDMI output is supported (although it’s super sluggish in 4k). During testing, I was able to get a 1080p display running at 144Hz without issue which felt pretty nice. The overall setup feels like a complete Raspberry Pi experience—ready to experiment with projects right out of the box.
Performance, however, leaves something to be desired. Running the Chromium browser at 1080p worked fine for light tasks, but the system bogged down on heavier pages with animation or video. At 4K, it felt even slower.
On the Speedometer browser benchmark, the Pi 500 Plus scored 4.54—less than half the score of an Intel N150 mini PC that costs the same or less. On the bright side, power draw was remarkably low: about 4 to 5 watts at idle and around 10 watts under load. That efficiency might make it appealing for someone experimenting with a home lab or lightweight Docker containers, but it’s not suited for media serving or anything that needs significant processing power.
The keyboard uses low-profile blue Gateron switches, and the keys are replaceable. The bright backlighting is nicely implemented, with adjustable colors and effects. There’s even a novelty game—Flappy Bird—that runs directly on the keyboard’s LEDs. The typing feel is solid, and having a dedicated power button is a welcome touch.
Overall, while the design is fun and nostalgic, the price puts it in direct competition with faster and more capable mini PCs. The keyboard is genuinely nice, but it drives up the cost of what’s still a modest performer. For light server duties or educational projects, it might find a niche, but the same money could buy more performance elsewhere.
I still like the charm of a computer built into a keyboard, but I’m hoping the next generation of the flagship Pi computer will push things further and correct some of the annoyances of the current generation. The Pi 500 Plus is a reminder of how far we’ve come since the $35 Raspberry Pi revolution—and how much the landscape has changed since then.
The FCC has taken an important step in its ongoing review of the ATSC 3.0 television standard, and this time, encryption—or DRM—has taken center stage. After months of advocacy, travel to Washington, and countless public comments, the agency has released a draft of proposed rulemaking that directly addresses many of the issues consumers have been raising.
The most immediate change in the document would eliminate the ATSC 1.0 simulcast requirement. Broadcasters using the new ATSC 3.0 system have been required to continue broadcasting the same programming on the older ATSC 1.0 standard to ensure that no viewers lose access. If this rule change is approved, stations could transition to the new format without maintaining a 1.0 signal, effectively speeding up the shift to next-generation TV. The FCC is also considering allowing MPEG-4 encoding on 1.0 channels to make more efficient use of spectrum, something that could benefit both broadcasters and viewers with modern televisions.
But the biggest topic is DRM and how it’s currently being managed. The FCC is asking for public comment on whether the commission should establish rules governing encryption of free, over-the-air broadcasts rather than leaving control to the A3SA, the private industry group currently setting those standards. The Commission expressed concern about how that system operates and whether it aligns with the Communications Act of 1934’s definition of broadcasting. They’re also examining whether DRM is creating unfair competition or restricting which devices can receive signals—an issue that has limited tuner availability and compatibility with platforms like Roku and Apple devices.
The filing also shows that consumer feedback has made an impact. Thousands of viewer comments have been cited throughout the document, and both Tyler the Antenna Man and I were mentioned multiple times. The FCC clearly took note of the frustration from early adopters who bought new tuners only to find themselves locked out by encryption controlled by a private organization. The agency even raised questions about fair use and whether DRM could undermine consumers’ rights to make in-home recordings—something the broadcast flag controversy of the ATSC 1.0 era had already tested in court two decades ago.
As the FCC prepares to vote on these proposals at the end of the month, it’s clear they’re not ready to sign off on encryption as it currently exists. The upcoming vote will likely remove the simulcasting rule, open another round of public comments, and extend the process into next year. That gives consumers and advocates another opportunity to weigh in. My hope is that the Commission will consider a real-world test by suspending DRM to see how the market responds. If encryption has been holding back tuner adoption and consumer interest, that experiment could prove it.
So we will likely need to keep submitting comments and addressing the questions the FCC is asking that relate to our own personal experiences. Tyler the Antennaman and I also suggest attaching selfies of “encryption error” screens that block free TV could help put faces to the issue. The FCC’s draft shows that public voices are being heard, and continuing to speak up might be what ensures free, open access to the airwaves remains part of the country’s broadcasting future.
The end of Windows 10 is coming up, with Microsoft planning to stop support on October 14, 2025. I’ve been seeing the same warnings you probably have — those pop-ups telling you to upgrade to Windows 11 — and I wanted to take a closer look at what that really means for people still using perfectly good older computers.
Windows 10 has had a long run, and I’ve always liked how well it performed even on lower-end hardware. The problem now is that Windows 11 has stricter requirements, mainly the need for a TPM 2.0 security chip and newer processors. If you’ve got an Intel 8th Gen or newer, or an AMD Ryzen 2000 or newer, you’re likely ok to upgrade.
Anything older isn’t officially supported, though there are ways around it. Microsoft doesn’t recommend circumventing the TPM chip requirement, and if they make a change assuming everyone has TPM 2.0, it could cause problems later. Business and government users also have to meet compliance standards, so running an unsupported version isn’t an option for them.
To see how this plays out in the real world, I fired up one of my older PCs — a small Shuttle box with a Celeron processor — and ran Microsoft’s PC Health Check app. It said I could upgrade for free, meaning this one squeaks by. Once Windows Update offers it, I can upgrade to Windows 11 in place. As always, it’s smart to back up first, but the process should be straightforward.
If your machine doesn’t qualify or you’re not ready to move on, Microsoft has something called the Windows 10 Extended Security Update (ESU) program. It’s available to consumers for another year, through October 2026. You can join it for free if you sync your PC settings with Microsoft, trade some Microsoft reward points, or pay $30. It’s not a long-term fix, but it buys more time for hardware that’s still working fine.
For people who’d rather try something new, Linux is worth a look. I tested Linux Mint on that same Shuttle PC, running the XFCE Edition since it’s lightweight and good for older systems. It’s surprisingly easy to get going, with a “live boot” option that lets you try it out without installing anything. Everything worked on my demo machine, and once installed, Mint has most of what you’d need — a web browser, office software, and access to more apps through its software manager. It uses about 1.2 GB of RAM sitting idle, so a 4 GB system runs comfortably.
Installing Linux does mean wiping the drive, so backups are essential, but if you’re done fighting with Windows upgrades, it’s a practical way to keep an older PC useful. I’ve noticed Linux often feels faster on aging machines than Windows 11 does, and since it’s supported well past 2029 for Mint’s current version, it’s a stable alternative.
Whether you stick with Windows 10 a bit longer, move to Windows 11, or jump to Linux, you’ve still got options. It’s interesting that after all these years, some of the oldest PCs still have life left in them — they just need a new OS to keep going.
I am working on a review of Lenovo’s Legion Go 2 (compensated affiliate link) and wanted to see how well it handled an external GPU before putting together a full review. The handheld includes two USB 4 ports that are Thunderbolt-compatible, so in theory, it should support external graphics. To test it, I used the GMKtec AD-GP1 — a compact eGPU enclosure that houses an AMD RX 7600MXT with 8GB of video memory. You can find the eGPU on Amazon here (compensated affiliate link). The AD-GP1 can power itself and the Legion Go 2 through a single cable, which makes for a cleaner setup.
The connection process was simple enough. I plugged the GMKtec into the top port of the Legion Go 2 and immediately saw it begin charging. Windows took a few minutes to detect everything, and I had to install AMD’s drivers for the external GPU. Once that was done, it recognized the card properly as a secondary GPU. For best performance, I disabled the Legion Go 2’s internal display and used only the monitor connected to the eGPU as pushing video through the handheld’s screen can reduce performance due to bandwidth limits.
Hot swapping between the handheld and eGPU worked on the Windows desktop, though not while gaming. You can plug and unplug the GPU while idle, but if you try to do it mid-game, it’ll crash.
Running 3DMark’s Time Spy benchmark, I saw a significant performance boost: the internal GPU scored about 3,999, while the eGPU setup hit 9,493. Power draw during the test hovered around 215 watts, leaving a little room within the AD-GP1’s 240-watt power budget.
Next, I tested No Man’s Sky at 4K with enhanced settings. The game ran smoothly most of the time, hitting around 60 frames per second with some dips. When I disconnected the eGPU mid-game, it predictably crashed, confirming that switching GPUs on the fly isn’t practical for gaming sessions. Still, for someone who wants a boost at home, an extra $500 or so for a dockable GPU setup delivers real gains.
I also tried other eGPU options, like the GT Box enclosure with a desktop RTX 4060 attached, but those setups are much bulkier. The GMKtec is smaller, cleaner, and more desk-friendly, even though it lacks extra USB ports for accessories. My only real complaint about the Legion Go 2 design is that one of its USB ports sits on the bottom, which makes docking awkward when the stand is open. You’d need an angled connector or stand to use it comfortably.
Despite its quirks, it’s satisfying to see a handheld like this handle desktop-class graphics with a single-cable setup. It’s not the most economical path to performance, but for those short on space who want one computer for both travel and desk use, it’s a workable approach. I’ll be putting together a full review of the Legion Go 2 soon and plan to focus on areas that haven’t been explored much yet — so if there’s something specific you’d like to see tested, let me know in the comments.
My latest Mini PC review takes a look at the Blackview MP100, a budget mini PC that is powered by AMD’s Ryzen 7430U — a six-core, twelve-thread chip with RDNA2 graphics. It’s not the newest processor out there, but it’s still plenty capable for light workloads and even some gaming.
The biggest limitation out of the box is that it only comes with a single stick of RAM, which means you lose out on dual-channel performance. That’s an easy fix, though: adding another stick of RAM dramatically improves its graphical capabilities.
When I opened it up, I found it surprisingly upgrade-friendly. It includes a 2.5-inch SATA bay for an extra hard drive, a spare NVMe slot, and upgradable RAM — you can go all the way up to 64 GB. My review unit had 16 GB in a single channel and a 512 GB NVMe SSD. There’s also decent cooling inside, which helps keep performance consistent under load.
The ports are practical: two USB-C connectors on the front (one of them full-function with DisplayPort output), USB-A ports, Ethernet, HDMI, DisplayPort, and a headphone jack. You can drive a 4K display at 144 Hz through DisplayPort, which makes it an interesting option for streaming via GeForce Now or other cloud-gaming services. The case itself is plastic, and it has RGB lighting on the top that can be customized or turned off through Blackview’s included software. That software uses a USB-to-serial driver for control, which looked a bit odd at first but turned out to be harmless after some malware testing.
Day-to-day tasks like browsing, watching 4K video, and general productivity felt smooth even with single-channel memory. YouTube playback at 4K 60 fps ran fine, and benchmarks put it in line with other mid-range Ryzen mini PCs.
Where the second RAM stick really matters is gaming and creative workloads. In No Man’s Sky, performance jumped from around 20–25 fps to 35–40 fps after installing dual-channel memory. 3DMark Time Spy scores told a similar story: 956 with one stick, 1,373 with two. Even emulation saw tangible benefits, particularly for PlayStation 2 titles.
Thermals and acoustics are well-managed for a small system. The fan isn’t loud during heavy use, though it can spin up unexpectedly when idle. Power draw hovered around 10–11 watts at rest and roughly 46 watts at full load.
Linux support was excellent — everything from networking to Bluetooth worked without issue with Ubuntu, making it a solid choice for running containers or self-hosted apps.
What stood out most was how close this system comes to being great with just a minor configuration change. Splitting that single 16 GB stick into two eights would’ve unlocked the full GPU potential without raising costs much. As it stands, the MP100 feels like a solid, efficient little computer with real potential once you make that one upgrade. It also carries a three-year warranty, which isn’t common in this price range.
Disclosure: Blackview sent the Mini PC free of charge. No other compensation was received, they did not review or approve the video or this post before publication and all opinions are my own.
In that earlier piece, I showed how Plex can fire off data whenever something happens on your server—whether it’s a play, pause, rating, or media being added. That data can then be sent to automation tools like N8N, which I’ve used in the past. But running an N8N server can be complicated and the hosted version requires a subscription.
In my latest sponsored Plex video, I demo a free, no-code path that doesn’t require hosting anything yourself using Make.com.
Make.com offers a free tier with 1,000 “credits” per month. Each credit equals a task, so something like receiving a webhook and then adding a row to a spreadsheet would count as two. For casual use, the free plan can go a long way.
I set up a simple workflow that starts by having Plex send a webhook to Make, which in turns takes some of that data and posts it to a Google sheet as a means of creating a log of watched content. That event is triggered anytime a new piece of media is played based on the state of the “event” tag in the webhook.
I then set up a second branch for when playback stops. In that case, I had the system call ChatGPT to generate recommendations for what to watch next, and then send that recommendation via email. The input for ChatGPT came directly from the webhook data, which made the recommendation specific to what I had just finished watching. It worked smoothly, and it showed how easily AI can be folded into these workflows.
For those thinking about linking Plex to home automation, the free, non-complicated options are a bit more limited. Services like IFTTT require a paid tier for webhook support, while open-source projects like Home Assistant can handle webhooks for free but require a bit more setup and complexity. Home Assistant, in particular, could be a strong option for linking Plex events with things like lights or thermostats, though it isn’t as straightforward as make.com.
There’s also Plex’s new official API, which was announced during Pro Week. The documentation is out, and I expect third-party developers will soon release connectors for platforms like make.com and N8N. That should make integration even easier, reducing the need to rely solely on webhooks. My YouTube colleague Techno Tim has already been experimenting with it, using the API to drive a dashboard for monitoring Plex server performance.
What I tried here was aimed at giving a taste of what can be done quickly without coding. Setting up webhooks with Plex and pairing them with make.com is accessible and flexible, especially for logging and basic automations. From there, it’s easy to imagine extending the idea into more complex routines, whether that means generating watch lists, building dashboards, or nudging your smart home into action when the credits roll.
I’ve been getting increasingly frustrated with how social media algorithms decide what to put in front of me. More often than not, what matters most to the platform are not my interests but rather content that the platform thinks will keep me glued to it.
Over the past few months, I started experimenting with self-hosted and hybrid solutions to build something I could actually control. What I ended up with is a little algorithm of my own that now emails me every morning with a curated digest of topics I care about.
The system runs on my Synology NAS using N8N, which I’ve also been using for other projects. The content engine pulls about 150 headlines a day from RSS feeds across trusted websites, YouTube channels, and Reddit forums I follow. From there, the workflow filters, organizes, and compiles the results into an HTML email.
It works really well. For example, when multiple outlets covered handheld gaming PCs, it was smart enough to recognize the GPD Win 5 and Asus ROG Ally as belonging to the same category and group them together. That gives me a cleaner view of what’s trending and helps me decide whether something is worth reviewing.
At the core of this is RSS, which has quietly persisted even as many sites moved away from it. I use TT-RSS to merge dozens of feeds into a consolidated source for each topic area. N8N then pulls those feeds into an AI agent workflow powered by Google Gemini’s free tier. I experimented with local models, but they couldn’t handle the complexity of parsing and structuring the data effectively. Cloud models still work better for this task, and because I only run it twice a day, I’m not paying anything for API usage.
Getting the prompt right was a big part of making this work. I had to iterate with both ChatGPT and Gemini until I landed on instructions that consistently returned useful results. The agent is told I’m a YouTube host looking for new topics, and I specify what types of content to prioritize and what to ignore. I also provide it with a structured HTML template so the output is consistent. The final email includes my calendar at the top, followed by curated sections on gadgets and cord cutting. It also uploads a copy to my FTP server so I can pull it up in a browser.
The advantage of this system is that I can fine-tune it. If something irrelevant slips in, I just add instructions to exclude it. If I want to emphasize a certain category, I can adjust the prompts. Unlike the opaque systems behind social platforms, this workflow only surfaces items from sources I choose and in the way I want to see them.
Everything I’m using—N8N, TTRSS, the Gemini free tier—is either free or open source. There are limits with the Gemini free tier, like rate caps and the possibility of data being used for training, but for my purposes it’s not a problem since I’m only working with publicly available content.
I haven’t put together an N8N installation tutorial yet, but Network Chuck has a good walkthrough that can help get N8N running on a server or NAS. It’s been interesting to see how popular N8N has become for building these AI agent tasks, and I’m trying to explore ways of using it that feel practical and useful. If you’ve also been frustrated by the way platforms filter your content, experimenting with something like this might give you back some control.
