FlashFeed

Full text

A Red Hat build engineer has ended his second experimental effort at using a high-end Arm64 desktop computer as his daily driver. His conclusions are instructive. Marcin Juszkiewicz’s series of posts on his tech blog about running a Fedora-powered beast of an Arm desktop computer has been interesting reading for about a year now. Almost exactly a year ago, he built a beast of an Arm workstation. He spent some €1,800 building a machine around an Ampere Altra, which as The Register described in 2020 is a serious bit of kit. And even at €1,800 (£1550, or $2000) quite a bit of it was second-hand, including the CPU and RAM. He described the machine’s capabilities in Arm desktop: 2025 attempt, part one. We were considering a story about the sheer cost of the kit, but his latest update describes The end of the AArch64 desktop experiment. That means it’s now more of a post mortem. To cut to the chase, he has switched back to using his AMD Ryzen 5 3600 system: the six cores of a $199 chip from 2019 outperforms an 80-core Arm64 powerhouse. Juszkiewicz, who also goes by the rather easier to spell hrw, knows his stuff, and he is a man of strong opinions. Indeed, his Mastodon profile gives a clue: “Tired of Seriously Bad Computers (SBC) so moved to Arm servers and virtualization.” SBC, of course, more normally stands for “Single Board Computers” in this industry, and by far the best-selling Arm SBC is the Raspberry Pi. As it happens, we’ve met him: the Reg FOSS desk attended a talk by Mr Juszkiewicz at the 2016 Flock to Fedora conference. We already knew that he wasn’t a fan of the Raspberry Pi devices. Also, for any doubters out there, this Altra box was not his first attempt to build an Arm64 daily-driver machine. That was back in 2015, as he described over three posts. AArch64 desktop: day one talked about the spec, involving a Mustang motherboard – which means an Applied Micro X-Gene X-C1. By day two, he started to cover issues with his distro: "I use Fedora on my machines. And as lot of people know this distribution has strange rules when it comes to multimedia support. Forget about MP3, H264 and few other things." The post talks about needing to build his own codecs and media player applets, the lack of Macromedia Flash support, and more. We note this just in case anyone is tempted to jump to the conclusion that his more recent Arm experiment failed due to a lack of knowledge or skills – he definitely has both in large amounts. After just four days, he wrote AArch64 desktop: last day, discussing the performance issues of an eight-core, circa 2.4 GHz ArmV8 computer: "OK, I can be spoiled by speed of my i7-2600k desktop but situation when Firefox with less than 20 tabs open is unable to display characters I type into textarea fast enough shows that something is wrong (16GB ram machine). And tell me that this is not typical desktop use of web browser…" He spares judging the machine too harshly: "I think that results may be affected by a fact that all I have here is Applied Micro Mustang based on X-Gene 1 cpu. It is one of first ARMv8 processors in Linux world and it is optimized for server use rather than desktop." Well, fast forward to 2025, and he still has no choice but to use a manycore Arm64 chip aimed at servers, rather than the desktop. As The Reg reported earlier this month, they are now a serious force in the server and datacentre market… but aside from Apple Silicon, not on the desktop or in laptops. This time around, in Arm desktop: 2025 attempt, part one, he starts by setting the context: "Almost ten years ago, I tried to use an Applied Micro Mustang as a desktop. And it was painful. " He also starts out with some of the problems. Even a decade on, he still has problems with missing codecs and client apps. The steps needed to work around them have changed, but the situation is not massively better. Last time, for instance, it was a problem that there was no Flash player; now, there’s no Arm Spotify client. In the second installment, Arm desktop: emulation he looks at the problems of using the latest x86-on-Arm emulation tools in trying to run x86 games on even a high-end Arm chip: "Test results were awful: Single core: 459 Multi core: 4110 That’s the level of an Intel Atom CPU from 2021." He concludes: “Let me be honest: without tweaking FEX-Emu config it was unplayable.” The third installment moves on to the issues of this server part: Arm desktop: so many cores, not enough speed. "Having 80 cores sounds nice, doesn’t it? But not so much during actual use… You see, building Fedora packages was flying by. With all cores in use, ccache buffers filling up (in case of rebuilds), and 128 GB of RAM in constant use, etc. "But at the same time, 100 percent load on all cores means you cannot listen to music on Spotify or watch online videos, etc. All that because the CPU cores are occupied by the build processes." His summary is short but not very sweet. The main issue: “The lack of single-thread speed.” And the bottom line: “To use a desktop system you do not need many cores. As long as they are fast.” In the final instalment, The end of the AArch64 desktop experiment, he looks at some of the issues in depth. He built his own custom specially patched kernels; he tried two different GPUs – both an AMD Radeon RX6700XT, a high-end card that was $479 when new in 2021, and then an Nvidia RTX 2060, a card that was $349 in 2019. With both cards, some apps wouldn’t run, due to missing OpenGL libraries, or they ran very badly: “Watching YouTube videos became impossible due to 720 out of 750 frames being dropped, etc.” He hasn’t pensioned off the machine: "The 'wooster' system stays powered on, churning through RISC-V package builds. It may be weak in single-thread, but it flies when it comes to multi-core load. " But this is the end for now: "As for the Ampere Altra, I am not planning to repeat this experiment. Another AArch64 desktop attempt would require a completely new hardware platform. And I have no plans to spend over twenty thousand PLN to buy an Nvidia DGX Spark system." (Twenty thousand Polish złoty was around $5,305, £4,020, or €4,660 at current exchange rates at the time of publication.) Our original plan when “hrw” started this series was to look into the lack of commodity-priced Arm64 hardware. At last year’s Ubuntu Summit, System76 presented its nearly ready COSMIC desktop, but it also demonstrated its 128-core Altra workstation, the formidable Thelio Astra. With half a terabyte of RAM and 40 TB of storage, it’s a snip at just over $6,000 (£4,600). We have to admit, after reading Hrw’s in-depth stories of what went wrong, we no longer covet one. Perhaps paying £168 for an 8 GB Pi 5 isn’t so bad after all. ®

Full text

One of the many millions of people captured by the Romans and sold into slavery was a man who died in Ravenna in the 1st century BC. He wrote his own story, on his own tombstone, so that everyone who saw it would know what he endured and what he survived.

Gaius Julius Mygdonius, Parthian in origin. Born a free man and captured as a boy, given away to Roman territory. When helpful fate made me a Roman citizen, I began to save for when I became 50 years old. Ever since my youth I have wanted to reach my old age. Now, stone, receive me willingly; with you I shall be released from my worries.

Gaius Julius Mygdonius wasn’t the name he was given by his father, and it wasn’t the name he bore when the Romans captured him, nor was it the name he was called on his travels to Italy or during his life in slavery. At some point he became merely Mygdonius to his captors and his enslavers, which simply denotes him as a person from Mygdonia (Bulgaria and northern Greece). Mygdonius describes himself as a Parthian, which means he was from ancient Persia.

We know that he was captured as a child after experiencing the Roman war, and possibly a Roman siege. Eyewitness accounts of the Roman war terrify even across a few thousand years of distance. When the Romans breached the walls of a city, they killed, indiscriminately, everyone they came across, following their orders to “kill all they came across, sparing none, and to not start pillaging until the signal was given.” Bodies filled the streets. Not just of humans, but of dogs and horses cut in half, says the Greek historian Polybius, who was himself taken captive in a Roman assault. The soldiers raped and killed until they were exhausted.

We have two eyewitness accounts of the process of being taken as a captive by the Romans. Neither is by someone who was himself enslaved, because both were too noble for slavery. They are non-Roman witnesses to Roman war. The first comes from the works of Polybius, a Greek statesman from the fantastically named city of Megalopolis who ended up first a captive in Rome following the Third Macedonian War in 168 BC, and then a tutor to Scipio Aemilianus. When Scipio grew up and became the leading general in the third war against Carthage, he took Polybius along with him — rather against Polybius’ will — and Polybius eventually wrote an account of what he saw during this time. The second account comes from Josephus, a Jewish scholar and leader born Yosef ben Matittyahu in Galilee who fought against the Romans in the Jewish War and was taken captive after the brutal siege of Yodfat in 67 AD. He accompanied the emperors Vespasian and Titus for the rest of the war until the sack of Jerusalem and the destruction of the Second Temple in 70 AD. Titus then took him back to Rome, where he developed a nice line in explaining Jewish history to the Romans and Roman history to the Jews.

Both Polybius and Josephus wrote accounts of the Roman process of taking and organizing their captives following a battle or siege and, although they were written 200 years apart, they are remarkably consistent. For a start, the life stories of both men provide useful examples that show that it was not local elites who were being enslaved by the Romans. Status protected people in captivity as it did in life, and no Macedonian or Bithynian leader ever ended up chained in a field in Sicily. The people who the Romans took captive and enslaved were the poor, the workers, and the already enslaved.

Mygdonius was one of these non-elites who survived the Roman onslaught. Still breathing when the signal came to stop the slaughter and start the pillage, Mygdonius became booty. Rounded up from his hiding place by a towering, battle-scarred soldier of Rome, he then waited to find out his fate. After hours, maybe days, of waiting the commander appeared to divide his captives; a tedious administrative job for a Roman general, but for Mygdonius and his fellow captives, perhaps the most important day of their lives. Would they be executed or enslaved? Would the general allow them to be ransomed, and was there anyone to ransom them?

Mygdonius was too lowly to be ransomed or taken into a relatively luxurious prison as a hostage, a fate open only to those blessed with noble blood. The choices facing him were death or slavery, and he was granted slavery. The Roman lawyers believed that slaves were called servi because generals saved ( servare ) them from death and so it was with Mygdonius. He was not given to the soldiers as a prize, or sent into public slavery, or kept as a treat by the general; he was simply sold there and then to the slave traders who followed the Roman army on even the smallest missions to take their booty and turn it into profit.

Both Polybius and Josephus offer detailed breakdowns of individual sieges and the aftermath in which the huddled, terrified inhabitants of cities were divided up into categories and sent off to new lives. Polybius describes the short and (relatively) uneventful siege of New Carthage (Cartagena, Spain) in 209 BC. Scipio took the city in a matter of days and without much struggle but still “as is the Roman custom” sent his troops to slaughter the population without mercy, dismembering not just the people but also their animals and burning their homes until they had the New Carthaginian general in custody. This worked pretty fast because, well, it would. Start cutting a person’s children and dogs in half in front of them and they do tend to despair.

Within a few hours Scipio had accepted the unconditional surrender of his New Carthaginian counterpart. The next day, he set to dividing up the booty, including the 10,000 prisoners. First, says Polybius, Scipio removed the political men with their wives and children and, asking them to look kindly on the Romans, he let them go free. This act of clemency was extended only to the city’s political elite.

The next group to be identified were the craftsmen or artisans, people like stonemasons and carpenters. These 2,000 men he made slaves of the Roman Republic, publicly owned to ply their trade for the good of their enemy. Then, Scipio picked out the strongest young men and condemned them to serve in the Roman navy as oarsmen. He also promised these men that if they helped him win the war and didn’t, y’know, die in hideous ancient naval battle or drown in the Mediterranean, he would free them when it was all over. All the rest of the prisoners, he handed over to the proper officials to be sold to the dealers who followed the armies. Polybius recorded these details because he believed Scipio to be uniquely merciful and moderate and to have generally demonstrated that the Romans were a great bunch of lads. The fates of the men, women, and children who were stripped, eyeballed, and sold on the auction block were of no more interest to him than the fates of the individual silver coins taken from the city’s treasury. They were merchandise now to be dealt with as property, not people.

As Cicero finished a minor campaign in Cilicia, he wrote a long letter to his bestie Atticus describing his adventures and finished up by saying that the only plunder he had kept for himself were the captives being sold at that moment. “As I write, I have 120,000 sesterces on the block,” he said, reducing all those poor people to cold cash and naked bodies.

The fates of the men, women, and children who were stripped, eyeballed, and sold on the auction block were of no more interest to [Scipio] than the fates of the individual silver coins taken from the city’s treasury.

Josephus describes pretty much the exact same process happening in Judea in 69 AD. There, the Roman generals were less interested in winning the goodwill of the population they had defeated and so Josephus claims that, in total, the Roman army took 97,000 people into slavery in the Jewish War, which is more people than can fit in Wembley Stadium. He describes Vespasian dividing up the captives at Tarichaea in Galilee, executing 1,200 of the elites, sending 6,000 of the strongest young men to dig Nero’s folly of a canal in Greece, and selling 30,400 people on the auction block.

These numbers blur. They are simply too big to conceptualize as individuals: 30,400 Mygdoniuses; 30,400 of your child, your mum, your brother; 30,400 human beings.

A year later, after the sack of Jerusalem, the Roman general decided to divide the captives up by age and looks: “He selected the tallest and most handsome of the youth and reserved them for the triumph; of the rest, those over 17 years of age he sent in chains to the works in Egypt, while multitudes were presented by Titus to the various provinces to be destroyed in the theaters by the sword or by wild beasts; those under seventeen were sold.” Very few of these would ever know freedom again; most would die hideously. Every one of them would be chained by the neck and ankle and wrist and led on nightmare journeys across deadly seas to foreign places. Every one of them would be beaten, at least as part of the process of making a free person realize that they were no longer a person, of turning a human into a slave.

Thus, with thousands of others, Mygdonius was chained at the neck and the leg, the cold metal resting heavily on his collarbone, chafing at his ankle, rubbing his skin raw and red. We see how enslaved people moved across the empire in the tombstone of Aulos Kapreilios Thimotheos, a dealer in slaves who had a line of eight chained men carved into his 2-meter-high memorial. This was the first step in Mygdonius’ transformation from a person into a slave, a life which hung between life and death. Maybe he thought about escaping, maybe he thought about suicide as he walked from the city to the sea, and then endured the Mediterranean crossing. Many, many men and women killed themselves on the boats to escape the violence and humiliation they knew lay ahead of them.

But Mygdonius always had hope. He says so himself. “Ever since I was young, I have wanted to reach my old age.” Captured as a child, Mygdonius clung to the dream of survival. Through his sale into imperial slavery, through the erasure of his old self and the creation of a new person: Mygdonius, the slave known only by his place of origin. Through his grinding years of enslavement, and his tireless work that finally led to his being freed. Through his third renaming, taking on Gaius Julius, the names of the people who enslaved him. Through it all, he remained a free Parthian in his heart. He saved and saved and worked and saved so that when he died, in his old age, his ashes could be interred in a huge sarcophagus of marble and the important facts of who he really was — free, freed, rejecting the slavery that sought to erase him — could be proclaimed forever to all who passed by.

This article The terrifying reality of being enslaved by the Roman army is featured on Big Think .

Full text

The famous opening scene of the Martian has Mark Watney stabbed in the torso with a communications antenna. While this accident sets up the plot for what is widely regarded as a modern classic of sci-fi storytelling, what if he was wearing a space suit that would have stopped the impact altogether? That’s the idea behind a recent NASA Small Business Innovation Research (SBIR) Phase I program run by researchers from Materials Research & Design, Fiber Materials, Inc, and NASA’s Johnson Space Center. Their work, which was recently presented at the National Space & Missile Materials Symposium, showcased a type of advanced 3D-reinforced fabric that could have saved Watney and his compatriots a whole lot of trouble.

Full text

Bitcoin Magazine

UAE-Based Goldman Lampe Private Bank Acquires $137 Million in Bitcoin

Goldman Lampe Private Bank has purchased €120 million (roughly $137 million) worth of Bitcoin, the UAE-based institution announced Monday, timing the buy to coincide with a recent pullback in cryptocurrency markets.

The Ras al Khaimah-headquartered bank said the acquisition strengthens its institutional Bitcoin holdings and reflects its conviction in digital assets as a long-term store of value. The move positions Goldman Lampe as one of the more aggressive institutional buyers to act on the current market downturn.

“Bitcoin continues to demonstrate remarkable resilience as a store of value and strategic asset,” said Abdullah Hamad Al Shamsi, Chairman of the Board. “By capitalizing on this market dip, we are not only enhancing our institutional holdings but also reaffirming our leadership in bridging traditional private banking with cryptocurrency solutions.”

Goldman Lampe did not disclose the exact number of Bitcoin acquired, the price at which the purchase was executed, or the bank’s total Bitcoin holdings to date.

Founded in 1934 and regulated in the UAE, Goldman Lampe markets itself as the first bank in the world to offer crypto term deposits — a product that lets high-net-worth clients earn yields on digital asset holdings within a regulated framework. The bank also offers gold bullion trading and private wealth management services.

The Bitcoin purchase adds to a broader thesis the bank has pushed publicly: that digital assets belong inside institutional portfolios, not alongside them as a speculative side bet. The bank’s term deposit product, it says, gives clients structured, compliant exposure to crypto in a format familiar to traditional wealth management clients.

The acquisition comes as institutional Bitcoin buying has become a more common playbook. Companies including MicroStrategy and a range of sovereign wealth vehicles have made similar dip-buying moves in recent cycles.

Bitcoin price action

Bitcoin entered June 2026 trading near $73,674 and has since fallen to around $58,500 as of this morning— a decline of roughly 18% for the month. The drawdown has been driven by ETF outflows, a stronger U.S. dollar, elevated interest rate expectations, and rotation into AI equities.

Technically, Bitcoin is trading below both its 20-month and 50-month exponential moving averages, signals that analysts associate with bearish intermediate-term pressure. The 50-day moving average sits above the current price, a potential resistance level if BTC attempts a recovery. The 100-month EMA, however, remains below the current price, leaving the long-term structural trend intact.

Bitcoin is down roughly $48,000 from its price one year ago, though that comparison reflects what was a period of peak pricing in mid-2025.

Goldman Lampe’s purchase puts its acquisition cost somewhere in the current trading range, with the bank betting that the dip represents an entry point rather than the beginning of a longer decline.

This post UAE-Based Goldman Lampe Private Bank Acquires $137 Million in Bitcoin first appeared on Bitcoin Magazine and is written by Micah Zimmerman .

Full text

Among many other things, Star Wars: Andor is widely regarded as the most visually impressive live-action Star Wars series to date. It's the perfect candidate for the sort of lavish art books Abrams Books does so well, and now you can add The Art of Star Wars: Andor (The Complete Series) to your collection.

With The Art of Star Wars: Andor in stores now, IGN can exclusively debut a spread from the book. Check it out below:

The Art of Star Wars: Andor (The Complete Series) is written by Lucasfilm's creative art manager Phil Szostak. This 288-page hardcover features no fewer than 500 pieces of artwork from the series, as well as interviews with star Diego Luna, creator Tony Gilroy, and many of the visual artists who worked on the series. The book also features a foreword by Tony Gilroy and an afterword by writer Dan Gilroy.

Abrams' official description of the book teases:

With deep dives into the locations that defined the series, including Kenari, Ferrix, Aldhani, Narkina 5, Mina-Rau, Ghorman, and Yavin 4, among many others, this essential volume celebrates the legacy of all twenty-four critically acclaimed episodes and the hurdles the filmmakers, cast, and crew had to overcome to bring their vision to life.

Every facet and detail of the Rebellion and Empire’s inner workings and the day-to-day existence within the world of Star Wars that Andor lovingly depicted are explored in great detail along with reflections on the series overall and its espionage roots and political relevance. The lived-in, battered costumes, ships, and locales that are so crucial to the DNA of Star Wars are on full display in this volume alongside the elegant formalism of the Galactic Senate and Coruscant.

The Art of Star Wars: Andor (The Complete Series) is priced at $60 and is available in stores now. You can order a copy on Amazon .

For more on the Star Wars franchise, check out IGN's review of Andor Season 2 and brush up on every Star Wars movie and series in development .

Interior spread from the upcoming book The Art of Star Wars: Andor (The Complete Series) (Abrams) by Phillip Szostak, on-sale June 30, available for pre-order now

© & TM 2026 LUCASFILM LTD.

Jesse is a mild-mannered staff writer for IGN. Allow him to lend a machete to your intellectual thicket by following @jschedeen on BlueSky .

Full text

As humans, we love repetition. Even if we know something is so bad for us, the fact that it's familiar is better than the scary unknown. So we do the same things over and over until one day we take that leap of faith and try something new. The bright side is that there's always the possibility for development; with that, life can never be boring. Philly’s Queen of Jeans have returned with their first new music since their great 2024 album All Again , with new single "Still Learning" to remind us that change has never been easy.

The post Queen Of Jeans – “Still Learning” appeared first on Stereogum .

Full text

Cards on the table, the first thing that piqued my interest about the Steam Machine's initial reveal were the interchangeable faceplates. After all, I am but a simple hardware creature with a penchant for the aesthetic. Unfortunately, Valve has no plans to release a whole slew of custom plates for their tiny game box.

We caught up with Valve engineers Yazan Aldehayyat and Pierre-Loup Griffais to talk all things Steam Machine , including customisation options. "Right now the two plates that come with the two terabyte SKU are what we're doing," Pierre-Loup explains, "And we don't really have any specific plans around what else [we might do]."

The 2TB Steam Machine comes with the 'red fabric and solid walnut' faceplates but, just because Valve has no plans to make more, that doesn't mean fresh faceplates are no-go from other sources, though. "We want to make sure that third parties can do everything they need to there, and that people tinkering at home can also do that," Pierre-Loup adds, "So, like usual, we'll be releasing all the CAD [3D files] and the dimensions for the outside, the attachment points for the magnets and the sizing of all that so that folks can make their own stuff."

Valve has already released the new Steam Controller's CAD files for easier customisation, though at time of writing we're still waiting on these files for the Steam Machine. That hasn't stopped some from taking matters into their own hands though.

Perhaps most notably, peripheral and accessory maker Dbrand teased custom Companion Cube housing for the Steam Machine last year . Unfortunately, they've had to start issuing refunds since the game box's release as it turns out the company had never actually got Valve's permission to make it.

To be clear, this was a third-party customisation being sold commercially, hence why Valve asked for it to be taken down. Fresh faceplates made for fun and not profit are unlikely to attract the same ire. So, yes, if you wanted to make your own Companion Cube inspired housing, you could—you just couldn't sell it like Dbrand attempted to.

I'm no 3D modelling wiz, so a single swappable faceplate appeals more to me anyway. Now, if you'll excuse me, I'm going to gather a bunch of cute Decoden bits, clear my schedule, and make my Steam Machine look like a tasty little snack (complete with fake whipped cream, my temps be damned).

Full text

According to reports, ChatGPT hit the one-billion-active-user mark in May. This makes it the fastest app to achieve this feat to date. I think this is enough for you to realize that it’s more important than ever to get your brand featured/ mentioned/ cited on ChatGPT and other LLMs online.

Unfortunately, this is not as straightforward, and before you start cooking up a new strategy to address this new era of search, you must understand where you stand with your brand right now.

This is where you start tracking how your brand is being perceived by these LLM models and how often you get picked over your competitors. This guide is here to answer exactly that.

For this, I tried a bunch of different methods to find the most economical and credible methods to track your business’s visibility in AI search results. Here’s everything that I found legitimate and actually working well.

What does AI visibility mean?

Before you pick a tool, you need to know which question you're answering. Most people merge three very different things into one word ("visibility") and then wonder why their data is a mess.

There are three questions, and they need three different methods:

Are you mentioned or cited in AI answers? When someone asks ChatGPT or Perplexity a question in your category, does your brand show up in the response? This is the one most people mean by "AI visibility."

Are people clicking through from AI tools to your site? A mention and a click are not the same thing. You can get named in a hundred answers and still get zero visits, because the user got what they needed and moved on.

Are AI bots even crawling your content? If LLM crawlers like GPTBot and PerplexityBot can't reach your pages, you're not eligible to be cited in the first place. This is the plumbing nobody checks.

Keep these separate in your head. A tool that nails question ONE tells you nothing about question TWO. I've watched smart marketers celebrate a rising "mention rate" while their actual AI traffic sat flat, because they were measuring two different things and treating them as one. Now, the methods.

Tracking AI visibility in AI search results: Free methods

Here is how you can get started with tracking your business in AI search results for free.

Google Search Console (GSC)

Google Search Console added a dedicated generative-AI report on June 3, 2026, and it's the freshest free tool you have. For the first time, it breaks down how often your pages appear inside AI Overviews, AI Mode, and the AI features in Discover, instead of burying that data in your overall numbers.

(Image credit: Future) This report measures the impressions, broken down by page, country, device, and date. So you can finally see which of your URLs Google's AI is actually pulling into answers, and watch that trend over time.

Now the honest caveats, because this isn't magic. It shows impressions only. There's no click data, click-through rate, or query data yet (Google says more is coming soon). The three AI surfaces are blended into one bucket, so you can't separate AI Overviews from AI Mode. And it's rolling out to a subset of sites, within the UK first, so it might not be in your account on the day you read this.

Check anyway. If it's there, set your baseline now.

A manual prompt log

The oldest method is still the cheapest, and it works on every platform. Write 10 to 30 questions your buyers actually ask, run them across ChatGPT, Gemini, Perplexity, and Claude on a fixed schedule, and log the results in a Google Sheet.

(Image credit: Future) Try to pick your prompts and then leave them alone. AI answers wobble run to run, so a single check is noise. Same prompt, same day next week, every week. The trend over a month is the signal, not any one result.

Is it tedious? Yes. Will it cost you a dollar? No.

Plus, it forces you to actually read the answers, which teaches you more about why you're getting left out than any dashboard ever will. If you only do one thing this week, do this.

Tracking AI visibility in AI search results: Custom workflow

If the manual log gets old (it will), you can automate it without paying for a full visibility platform using a workflow automation tool like n8n .

It is a workflow automation tool, and its template library has ready-made workflows that do exactly what the manual log does on autopilot. The pattern is that it reads your prompts from a Google Sheet, sends each one to ChatGPT, Gemini, and Perplexity, has a step check for every answer for your brand, and writes the results back to your sheet on a schedule.

Browse n8n's template search for "brand visibility" and you'll find a few versions to start from.

(Image credit: Future) The cost for this process is genuinely small. You bring your own API keys, and on cheap models, the OpenAI and Gemini calls run for a pretty small amount. Perplexity is the main line item, because its Sonar API adds a per-request search fee of roughly $5 to $14 per 1,000 requests on top of tokens. For a 25-prompt list run weekly, you're looking at low digits per month.

Tracking AI visibility in AI search results: Using A paid tool

If you'd rather not build anything, a purpose-made tool collapses all of the above into one screen.

Getting these tools set up is generally fast. Typically, you start by entering some brand information such as your website, name, and industry. The tool will then work on your behalf, returning insightful information such as:

Your "share of voice" across different AI platforms

Performance breakdown based on model (such as ChatGPT, Gemini, and Claude)

Which websites and brands are being referenced where you want to be

Gaps where competitors aren't mentioned

Grading your website content for AI readiness

When is paying worth it? When you're tracking this across several brands, or you need a report someone else will read, or your time is worth more than the subscription.

For a single brand on a tight budget, the free tier plus a prompt log will take you a long way first. There's no shame in starting free and upgrading when the manual work starts eating your week.

So which one should you actually use?

Start with where you are. If you have no budget, turn on the Google Search Console AI report today, then run a manual prompt log every week. That combination answers all three questions: who Google's AI surfaces, who clicks through, and who gets mentioned across the chatbots.

If the manual log gets painful, automate it with an n8n workflow for the chatbots. You'll spend a few dollars a month and save yourself hours.

If you're doing this for more than one brand, or you need clean reports without the building, a subscription tool earns its keep.

One last thing, because it's the mistake I see most. Don't track AI visibility once and call it done. The whole reason the timestamp matters, the whole reason you sample weekly, is that AI answers change constantly.

The number you pull today is a snapshot. The line you build over a month is the actual story. The brands building that baseline now are the ones who'll know exactly where they stand when AI search becomes the front door to everything.

Frequently Asked Questions

Which AI platform should I focus on first?

Start where your buyers already are. ChatGPT is the obvious first stop, since it has more users than every other assistant combined. If you already pull traffic from Google, watch AI Overviews and AI Mode too, because they ride on the search presence you've already built. And if you sell B2B or anything research-heavy, keep an eye on Perplexity, since that's where the careful, comparison-shopping buyers tend to be.

How many prompts should I track, and how do I pick them?

10 to 30, and how you pick matters more than how many. Cover three types: direct brand questions ("Is [brand] any good?"), category and comparison questions ("best [thing] for [use case]") and the problem questions people ask before they've heard of you ("how do I fix [problem]"). That last bucket is where most brands find their biggest gaps. Use the questions your buyers actually type, not the polished ones you wish they typed.

If people read the AI answer and never click, does being mentioned even matter?

Yes, and arguably more than a click ever did. When an AI names you as the answer, you're the recommended option at the exact moment someone's deciding, which is the most valuable spot in the entire buying decision. The click is a bonus, not the point. It's the difference between a friend recommending you by name and a friend handing over a list of links. You want to be the name.

Is AI visibility the same as SEO? And does SEO still matter?

They're cousins, not twins.

SEO is about ranking a blue link that someone clicks. AI visibility (you'll hear it called GEO or AEO) is about getting named or cited inside the answer itself. They overlap a lot, especially on Perplexity and Google's AI, which both lean on pages that already rank.

So no, SEO isn't dead; it's often the foundation on which the AI citation is built. But you can sit at #1 on Google and still be missing from the AI answer sitting above it, which is exactly why you track the two separately.