Is Cadillac’s Lyriq EV an exception to the rule, or a sign of things to come?
Car manufacturers are turning away from Android Auto and Apple CarPlay in favor of native infotainment systems. The trend has generated a fair bit of controversy, with critics labeling it as an industry-wide pivot toward subscription models and data-related cash grabs. Executives, meanwhile, have billed the move as providing a more seamless, functional in-car experience.
One of the companies most invested in this move is General Motors. Last year, CEO Mary Barra confirmed that GM would phase out Apple CarPlay and Android Auto from its gas-powered vehicle lineup, doubling down on the controversial decision to drop the popular entertainment systems from its EVs in 2023. In its place, the company is upgrading its in-house infotainment, replacing its built-in Google operating system with Google’s AI assistant, Gemini. In an April press release, the company stated that it hopes to deliver “a custom-built AI assistant, fine-tuned with proprietary vehicle data to help anticipate your every need” by the end of 2026.
Advertisement
However, not every vehicle in GM’s lineup will be losing access to third-party infotainment systems. Reportedly, the 2027 model of Cadillac’s luxury EV, the Lyriq, will continue to support both Android Auto and Apple CarPlay. The decision to preserve the Lyriq’s access to phone mirroring systems is surprising, given GM’s recent transition toward native applications in EVs like the Chevy Silverado EV Work Truck and GMC Hummer EV, both of which saw their CarPlay and Android Auto functionalities removed earlier this year. As it stands, the electric SUV will be the only GM EV sold in North America compatible with both infotainment systems.
Exception or the rule?
While GM’s transition from Apple CarPlay and Android Auto might be controversial, company executives have stated that it will ultimately lead to a stronger product. In an interview with Nilay Patel on The Verge’s Decoder podcast, Barra stated that the company had received customer feedback that “it was very clunky” moving between GM’s in-house system and third-party applications. Now, GM is banking on customers preferring an AI-driven system with a host of verbal commands and smart features, ranging from smart route planning and curated entertainment systems to hands-free texting and streaming applications.
Advertisement
With this in mind, GM’s logic behind the Lyriq is puzzling. In a CarBuzz report on the luxury EV’s infotainment options, a Cadillac representative explained that the company preserved the Lyriq’s access to phone mirroring applications “based on customer feedback.” However, it’s unclear why this logic has not applied to GM’s other models, or whether it is a signal of moves to come.
In prioritizing its own application, GM has made efforts to grant access to key non-native applications — a commonly cited impetus for consumers hesitant to migrate to their car’s in-house infotainment system. In December of last year, for instance, GM announced that Apple Music would be available on its in-car software platform for all 2025 and 2026 models. Tim Twerdahl, GM’s VP of global product management, touted the addition as “the latest example of how we’re expanding entertainment choices built directly into our vehicles.” With its new Gemini-powered system, GM’s Google compatibility gives you access to Google Maps and other popular applications available on the Google Play Store.
For drivers of non-Lyriq GM EVs who prefer their phone mirroring application, the EV Play LT is a nifty workaround. Capable of providing wired and wireless phone mirroring for Chevy, GMC and Cadillac EVs, the EV Play LT is a third-party product that will deliver Apple CarPlay and Android Auto functionalities in infotainment systems without the popular phone mirroring apps. At $199.99 and with no subscription fees, the product might be just what the infotainment doctor ordered.
The acquisition was expected to help Anthropic’s compute infrastructure absorb its growing demand.
Anthropic is backing away from talks to acquire Decart AI, a research start-up building technology that improves chip efficiency, Bloomberg news reported earlier today (8 September).
The deal was expected to be worth around $6bn, significantly higher than Decart’s reported nearly $4bn valuation following a $300m raise earlier this year.
Backed by Nvidia, Benchmark, Adobe, Sequoia and Radical Ventures among others, Decart’s AI infrastructure improves training and inferencing capacities of AI models.
Advertisement
Its foundational model Lucy supposedly edits live video at the speed of stream, while its so-called ‘world model’ Oasis, designed for physical AI, generates realistic simulations to train robots.
The acquisition was expected to help Anthropic’s compute infrastructure absorb some of its growing demand. The AI giant, according to The Information, has signed some $517bn in compute capacity leases in just over the past 11 months, which amounts to some 14.8GW of capacity.
The company also confirmed last month that it would be designing its own chips to circumvent the worldwide shortage in the technology and access a steady supply.
Headquartered in California, Decart was founded in 2023 by former Israeli Defence Forces’ intelligence unit soldiers; brothers Dean and Orian Leitersdorf and Moshe Shalev. Its $300m May raise was led by Radical Ventures and its co-founder Jordan Jacobs, and brought Decart’s total raise to more than $450m.
Advertisement
Reports of the failed Decart acquisition come as Anthropic is expected to delay marketing its initial public offering prospectus to mid-October. The AI giant’s highly anticipated public listing, investors hope, will make the largest debut in history, raising the company $75bn or more at a $2trn valuation.
A Clinton Township, Michigan man was charged $418 for a single EV charging session at Elder Hyundai’s Macomb Township store after finding the charger listed as publicly available on the EVgo app, but didn’t discover the charges until after wrapping up his 40-ish minute charging session. “I was flabbergasted,” explained Stanley Kpenkann, who had been driving a rented Chevy Bolt EV at the time. “Never in my wildest dreams did I think I would be saddled with a bill of $418 for a charging session.”
After calling EVgo to report the error, he found that the exorbitant charge wasn’t a mistake at all – Elder Hyundai was charging customers an eye-gouging $5 per kWh, plus another $5 for every minute connected.
A local news report notes that “After back-and-forth with EVgo, the company processed what it called a goodwill refund of $405.” (And the dealership argued their charger was reserved for its own customers and vehicles, so they set prices high to deter outside use.)
But Electrek argues the story “speaks to much larger problems among the EV charging business in particular and car dealerships in general…”
Advertisement
[In June] Plug & Play EV’s Steve Birkett took to LinkedIn to call out his local Hyundai dealer in Union, NJ for charging a ridiculous $15 per kWh to top off his IONIQ 5 (!)… Even worse, he noted that a number of area dealerships that had listed their ports as “public” were actively discouraging public use with signs and stickers designed to turn people away…
[U]nlike gas stations, EV charge point operators in many states aren’t required to clearly advertise what they’ll charge before a driver plugs in in the same way that gas stations are: with clearly legible signage that’s visible from the street.
Switzerland has a population of 9,154,242. And its federal government just launched a pilot program to test open source alternatives to Microsoft 365 on 3,000 workstations, reports the blog It’s FOSS. “That’s about 7% of the federal workforce.”
The target is to complete the migration by end of 2027… On September 3, 2026, the Federal Council published results of “PoC BOSS”, a feasibility proof-of-concept involving 172 federal employees who tested the openDesk suite, a German open-source collaboration platform. During the proof-of-concept phase, core office tasks like document processing and email received positive assessments, while large-scale video conferencing still showed technical limitations. Based on the ‘success’ of the PoC phase with 172 employees, the pilot is now launched for 3,000 employees…
According to Matthias Stürmer, professor at the Bern University of Applied Sciences, Microsoft’s supremacy in public institutions poses three problems that are driving this migration. First is the risk of foreign access. US cloud legislation could expose Swiss government data to foreign authorities. Second is the risk to service continuity, as dependency on a single foreign vendor creates operational risk. The third risk is the escalating costs as proprietary licensing fees are rising with no Swiss leverage.
Switzerland’s military cybersecurity unit, Cyber Command, is not waiting for the civilian pilot. It is already poised to replace Microsoft 365 entirely with openDesk by October 2026. It is pretty much the same reason. Military doesn’t want foreign governments accessing sensitive Swiss data.
The article notes that during the pilot phase, the new system runs in parallel with Microsoft 365 rather than replacing it. But the article’s author speculates that “If the pilot is successful, we might expect the migration to continue on all the 54,000 workstations owned by the federal administrations.”
Carrying culinary adventures into the unknown comes the Le Creuset x Star Trek: Galileo shuttlecraft casserole dish ($150), which has a roomy 1.6 liters of space inside. Vulcan votaries are well catered for with the $35 Spock ear mug (perfect for Vulcan spice tea), complete with gold detailing along the rim as a nod to the Starfleet uniform; the $55 Spock hand spoon rest; and the whimsical Vulcan salute “live long and prosper” oven mitt, resplendent in signature Star Trek baby blue for $40.
Not merely content with digitally launching the collection online, Le Creuset engaged Sent Into Space to boldly send its Enterprise casserole dish where no pan has gone before—more than 111,000 feet above the Earth, atop 99.5 percent of the planet’s atmosphere. While obviously a blatant (albeit fitting) marketing stunt, there’s actually some considerable logistics to overcome in sending a cast-iron pot to the edge of space.
United Federation of Pan-ets: the Star Trek Enterprise casserole dish at 111,000 feet above the Earth.
Courtesy of Le Creuset
Sent Into Space co-founder Chris Rose had the task of launching the pot in the stratosphere via a space-capable balloon taller than the height of The White House, where it stayed for 1 hour, 19 minutes, and 4 seconds enduring temperatures down to -63.4°C (-82.12°F). “Because of the weight, we had to give some serious consideration to how we were going to hold things, making sure everything was very balanced,” Rose tells WIRED. Rose and his team also have to rely on simulations to take into account considerations such as where will the pot land as it’s parachuted back down to Earth, as well as how to avoid bodies of water and built-up areas, for example.
Advertisement
“The legal authorities within the aviation community need to know. We can’t have a Le Creuset casserole dish floating past the window of an airborne A380 as it makes its way off to some destination,” Rose says. “So we are in constant communication, both from a flight-clearance perspective before this ever leaves leaves the ground, to onboard systems for telemetry and communication.” Indeed, in the early stages of reentry, Rose says the Enterprise casserole dish reached speeds in excess of 300 mph before slowing and landing safely in Lincolnshire, England.
Back on terra firma, the Le Creuset x Star Trek collection should serve cadets and captains alike, provided you can secure your selections before stock vanishes like a cloaked Bird-of-Prey.
The year 2026 marked the arrival of a handful of new acronyms in the TV world, with regular old mini-LED now having to share space with SQD, RGB, Micro RGB, True RGB, and other related terms. While those last three are basically marketing labels for the same RGB-backlit LCD display technology, SQD was coined by TCL and stands for Super Quantum Dots. SQD is something entirely different from RGB tech, though it is mostly an evolution of the Quantum Dot technology long used in mini-LED TVs.
So far in 2026, eCoustics has reviewed two of TCL’s SQD Mini-LED TVs, including their flagship X11L, and their step-down QM8L, along with rivals featuring RGB Mini-LED TV technology from Samsung, Sony, and Hisense. Of these, the TCL QM8L and Hisense UR9, the company’s flagship RGB Mini-LED TV, are often similarly priced, especially at the 65-inch size with both currently selling for under $2,000. How do these two top Mini-LED TVs stack up, and which model is worth your hard-earned money? Let’s compare them side by side.
SQD Mini-LED vs RGB Mini-LED: What’s the Difference?
The Hisense UR9 and TCL QM8L are both Mini-LED-lit LCD TVs. Where they mainly differ is in the design approach of the TV’s backlighting system.
RGB Mini-LED TVs like the Hisense UR9 use micro-sized red, green, and blue LEDs in place of the blue or white LEDs found in mini-LED TV backlights. RGB backlighting can, in theory, at least, provide more accurate color reproduction than standard LED backlights, and it can also deliver enhanced brightness and color saturation, along with extended BT.2020 color space coverage.
Advertisement
Super Quantum Dots are essentially a refined version of standard Quantum Dot technology. In SQD Mini-LED TVs like the TCL QM8L, the SQD layer is paired with an Advanced Color Purity algorithm and an Ultra Color filter that uses 5-nanometer particles instead of the 60-nanometer particles that are standard for Mini-LED TVs. The end result, according to TCL, is more efficient and accurate pixel-level color transmission, along with extended BT.2020 color gamut coverage.
The Hisense UR9 and TCL QM8L both feature the Gemini for TV AI assistant
Hisense UR9 vs TCL QM8L: Features
The Hisense UR9 is available in 65-, 75-, 85-, and 100-inch screen sizes. It uses the company’s Hi-View AI Engine RGB processor, which provides multiple AI-based picture enhancements, including a Personalized Customization feature that guides you through the process of creating your own picture mode. An “Obsidian Panel” advanced low-reflection screen surface helps to reduce screen glare on UR9 series TVs, and they feature Dolby Vision IQ, HDR10+ Adaptive and Filmmaker Mode picture presets. A software upgrade scheduled for late 2026 will add Dolby Vision 2 Max to the UR9’s feature set. As this comparison was going live, we just received the Dolby Vision 2 update on our review sample, so stay tuned for more details on that in an upcoming article.
The TCL QM8L is sold in 65-, 75-, 85-, and 98-inch sizes and uses a high-contrast WHVA 2.0 Ultra Panel with a wide color viewing angle feature and an anti-reflective screen filter. TCL’s TSR AI Pro processor brings AI enhancements for contrast, color, motion handling, and upscaling. HDR format support includes Dolby Vision, HDR10+, HDR10, and HLG, and there are Filmmaker Mode and IMAX Enhanced picture presets. Like the Hisense UR9 series, the QM8L series is slated to receive a Dolby Vision 2 Max upgrade via a software update scheduled for late 2026.
Both TVs come with an adjustable-height pedestal stand (TCL QM8L in photo).
Hisense UR9 vs TCL QM8L: Design
Hisense’s RGB Mini-LED TV has a slim bezel but the panel thickness approaches two inches, giving it a relatively bulky look. One advantage to the TV’s thick panel is it provides ample room for a powerful built-in 80W, 4.1.2-channel speaker system, which has been tuned by French audio manufacturer Devialet and supports both Dolby Atmos and DTS:X. An included aluminum pedestal stand has an adjustable height option to clear space for a soundbar, and Hisense’s full-featured remote control features a backlit keypad.
Three HDMI 2.1 ports are located on the UR9’s back panel, along with an antenna input for the TV’s ATSC 3.0 tuner. Curiously, there is also a DisplayPort-over-USB-C input, an unusual feature that lets you connect a computer, tablet, or phone directly to the TV’s bottom-left edge.
TCL’s SQD Mini-LED TV has a ZeroBorder design that gives it an appealing “all-screen” look, and while panel thickness tops out at two inches, the TV’s sides taper gently to create the illusion of thinness. The adjustable-height aluminum pedestal stand can be elevated or set flush with a TV stand surface, and TCL’s new, full-featured remote control has a backlit keypad. Connections on the QM8L include four HDMI 2.1 ports and an antenna input for the TV’s ATSC 3.0 tuner.
Advertisement
The QM8L has an Audio by Bang & Olufsen speaker array that uses down-firing speakers on the TV’s bottom surface combined with two “subwoofers” located on the back. The quality of the TV’s built-in sound is good overall, but you may want to instead take advantage of its Dolby Atmos FlexConnect support, which lets you pair up to four wireless TCL satellite speakers and a subwoofer for a 4.1.4-channel system.
Advertisement. Scroll to continue reading.
Hisense UR9TCL QM8L
Hisense UR9 vs TCL QM8L: Picture Quality
The Hisense UR9 had excellent brightness results when we tested it using Portrait Display’s Calman Color Calibration software, topping out at 3,345 nits in Filmmaker Mode for HDR and a whopping 1,698 nits for SDR. Color results were also very impressive, with the UR9 managing to cover 91.5% of the BT.2020 color gamut and 99.4% of DCI-P3. The TV’s measured color accuracy in Filmmaker Mode was also very good out of the box.
Numbers don’t always tell the full story, but the Hisense’s bright picture was great for daytime sports viewing, and movies with a rich color palette also came across as punchy and clean-looking. The UR9 performed well with reference “dark” movies, showing strong shadow depth and detail. Our main complaint from a picture quality standpoint was the TV’s off-axis picture uniformity, with both color and contrast fading noticeably when viewed from off-center seats.
The TCL QM8L also had high brightness, peaking at around 4,500 nits on some measurements, though a more realistic expectation would be around 2,000 nits in Filmmaker Mode for HDR. Color measurements slotted in a bit below the Hisense UR9 at 89.3 for the BT.2020 color gamut and 97.7 for DCI-P3. Color accuracy was a bit disappointing at an averaged 4.9 for grayscale and 3.3 for color, but both results could be easily improved through calibration.
Advertisement
Like the Hisense, the TCL’s high brightness makes it great for sports viewing. The QM8L also benefits from a wide viewing angle, with color maintaining saturation even at far off-center seats. Also, its anti-reflective coating is effective at reducing screen glare in bright viewing conditions. Colors in movies look rich, but not exaggerated, and dark movies benefit from the TV’s strong contrast, refined shadow detail and almost non-existent backlight blooming artifacts.
The Google TV Home pageGoogle TV Live TV guide
Hisense UR9 vs TCL QM8L: Smart TV
Both the Hisense UR9 and TCL QM8L use the Google TV platform for streaming and smart home control. Google TV has a relatively streamlined interface, and it can present customized program recommendations based on your viewing and web-browsing history when linked to a Google account. The version of Google TV found in both TVs is enhanced with the Gemini AI assistant, which lets you do everything from researching your next vacation to generating custom screensavers using voice commands (the Hisense and TCL feature a far-field mic option along with a built-in mic on the remote control). Both TVs additionally feature a Live TV grid guide for browsing Freeplay free ad-supported streaming channels and broadcast channels pulled in by the set’s ATSC 3.0 tuner.
The Hisense UR9 has three HDMI 2.1 portsFour HDMI 2.1 ports are provided on the TCL QM8L
Hisense UR9 vs TCL QM8L: Gaming
The Hisense UR9’s strong suite of gaming features includes a native 180Hz refresh rate, FreeSync Premium Pro VRR, ALLM and Dolby Vision gaming. HDMI 2.1 ports are limited to three, however, which can be a limiting factor for gamers with multiple consoles. Input lag measured using a Bodnar 4K meter clocked in at 12ms in Game mode on the UR9, a result below the 16ms threshold that serious gamers generally consider acceptable.
The TCL QM8L’s gaming features and performance are both a step above what we’ve found in previous TVs from the company. It supports a 144Hz refresh rate, along with 1080p/288Hz for PC gaming, and there’s also FreeSync Premium Pro, ALLM, and Dolby Vision gaming. Input lag measured 9.8ms for a 4K 60Hz input in our testing, a result that slightly bests the Hisense UR9 and is comparable to the best gaming TVs from LG and Samsung.
Hisense UR9 vs TCL QM8L: Value
At $1,899.99 for the Hisense UR9 and $1,399.99 for the TCL QM8L (both prices for a 65-inch model), each of these next-gen Mini-LED TVs is a strong value. Both offer high brightness, refined local dimming and extended color gamut coverage. And both can be upgraded for Dolby Vision 2 Max via software updates, making them among the few TVs to support that exciting new feature.
Compared to other RGB mini-LED TVs such as the Samsung R95H ($2,999.99 for the 65-inch model), the Hisense offers notably good value. It lacks some of that TV’s key features such as a matte finish Glare Free screen, and Samsung’s Tizen is a superior smart interface compared to Google TV, but it’s still tough to overlook that $1,000-plus price differential.
Advertisement
The TCL QM8L with its Super Quantum Dots tech is a different animal altogether from the Hisense and Samsung, but its measured performance makes it a close match for both, and we found its picture quality to be exceptional when we tested it. At $1,399.99, it’s a steal.
Hisense UR9TCL QM8L
The Bottom Line
Both of these sets count among the best TV values you can find today, offering great overall picture quality with exceptional brightness and cutting-edge features such as Dolby Vision 2 and Gemini AI support. Both are also impressive gaming TVs.
Making a close comparison of the two, the Hisense has better measured performance, besting the TCL when it comes to BT.2020 and DCI-P3 color gamut coverage and out-of-box color accuracy. But in subjective tests, we found the TCL’s picture quality to be equally impressive, and its off-axis uniformity and screen glare rejection in bright rooms were both superior to those of the Hisense.
Advertisement. Scroll to continue reading.
To cut to the chase, the Hisense UR9 is a great value for an RGB Mini-LED TV, but the TCL QM8L is a great TV value, period.
Finished on a new Guangzhou line, XPENG’s Iron stood up and walked away on its own. No remote pilot. No stage crew. Just a completed humanoid leaving the factory floor the way a car leaves a paint booth, only this one used two legs. The company called the September 7 moment the first time an advanced general-purpose humanoid had been built on automated production lines and then departed under its own power.
XPENG claimed the factory as the first fully automated line constructed for humanoid robots of its kind. More than 80% of the production line’s fundamental steps have to be automated. They used their existing quality systems from e-vehicle factories, but said they needed to be altered to suit the tighter tolerances required for humanoid robots. The goal was to build a production line that could consistently produce high-quality products and scale up without tearing down the entire facility. He Xiaopeng, chairman and CEO of XPENG, stated that when they began to develop the lines, they had no prior playbook to follow. He described the success as a little first step toward opening up an entirely new product category.
BUILD, CODE & DRIVE YOUR OWN ROBOT CAR: Turn coding, electronics and engineering into a working programmable robot car you can assemble, program and…
EXPLORE FPV, LINE TRACKING & OBSTACLE AVOIDANCE: Control the robot with the ELEGOO app or IR remote, view live FPV video through the onboard camera…
BEGINNER-FRIENDLY BUILD WITH GUIDED WIRING: Keyed XH2.54 connectors help reduce wiring mistakes, while the illustrated tutorial and example programs…
Iron, the humanoid robot, has a remarkable 76 degrees of flexibility in its body and an additional 21 in each hand. It has a completely enclosed, flexible lattice construction that not only gives it a human-like appearance but also has a sophisticated safety net design. So it can now survive the rigors of manufacturing with greater security. Three custom-designed Turing chips deliver an impressive 2,250 TOPS (tera operations per second), more than enough to operate the company’s Physical AI base model on the robot itself rather on a remote server. On paper, this should reduce latency and allow the device to do complex tasks without requiring human control via a joystick. Officially, it is a platform designed to handle a wide range of tasks, but it can also continue to learn as it moves around the world. It will begin with a low-key distribution of tasks for XPENG in its own stores and campuses later this year. In a few months, it will be on the move in China and over the world by 2027.
Iron is a large machine, standing 178 cm (5.8 ft) tall and weighing 70 kilograms (154 pounds). It has 76 degrees of freedom in its body, with an additional 21 in each hand. A flexible lattice framework wraps around the entire object, giving it a human appearance while yet meeting safety standards. Three unique Turing-type AI chips give the power to operate the company’s Physical AI base model directly on the robot. This should keep things relatively nearby, resulting in reduced latency and eliminating the requirement for a human to be trapped holding a controller someplace in a faraway location.
Battery-wise, they are putting high-energy solid-liquid hybrid batteries into production. By the end of 2026, XPENG will be churning out IRON by the hundreds. Following that, the first batch of robots will begin to work inside business stores and campuses in 2027, followed by a wider rollout in China and other markets that same year. If all goes as planned, they hope to be producing over 1,000 units each month by the end of 2026, with the much longer-term objective of selling a million units by 2030.
NeoSapien is working to make it easier for people to capture and remember important conversations. The company began this approach with Neo 1, which we have reviewed, a wearable that could listen to, understand, and remember conversations in the background. Now, NeoSapien is expanding its AI memory ecosystem with its Desktop App and NeoSapien MCP. While the Desktop App captures conversations across digital and physical settings, MCP brings that context into AI tools such as ChatGPT, Claude, and Cursor.
NeoSapien Desktop App Captures Conversations
The NeoSapien Desktop App captures meeting audio directly from the computer without relying on bots or platform-specific integrations. It works with Zoom, Microsoft Teams, Google Meet, WhatsApp, and other desktop-based platforms. The app runs in the background, allowing users to continue with their existing workflows. It can also capture in-person conversations through the computer’s microphone. The app automatically creates transcripts, AI-generated summaries, and searchable notes from captured meetings. Its calendar integration shows upcoming meetings and lets users start capturing them with a single click.
The Desktop App also supports multilingual conversations where speakers switch between Telugu, Kannada, English, Spanish, and other languages. NeoSapien’s AI engine understands context across more than 100 languages, including Kannada and Mandarin. This makes the app useful for Indian businesses and global teams working across different regions and languages.
NeoSapien MCP Brings Personal Context to AI Tools
NeoSapien MCP extends the company’s AI memory system beyond the Neo device and Desktop App. It lets users bring captured conversations and memories into AI platforms such as ChatGPT, Claude, and Cursor. This removes the need to copy and paste information or manually switch between apps. For example, sales teams can use client conversations to create structured proposals, while developers can use planning discussions inside Cursor. Teams can also turn meetings into summaries, follow-ups, and action items.
The technology separates the process of memory capture from the location where users decide to use the captured memory. Consequently, crucial contextual information will be able to migrate between various AI applications rather than being limited to a single application. This strategy has been termed the personal memory layer of the AI ecosystem by NeoSapien. NeoSapien refers to this concept as a personal memory layer for the entire AI ecosystem.
Advertisement
NeoSapien’s Vision for Connected AI Memory
The NeoSapien Desktop App and MCP work together to build a broader AI memory ecosystem. The Desktop App captures conversations from both physical and digital environments, while MCP makes that context useful across the AI tools users already rely on. This approach helps users stay focused on their work while keeping important conversations, decisions, and follow-ups accessible when needed. NeoSapien aims to make personal memory portable across platforms, allowing AI tools to retain relevant context and provide more connected and useful assistance in everyday work.
“A team at Google’s research arm used Google Maps to reroute a small portion of drivers in 10 U.S. cities,” reports SFGate — including San Francisco and Los Angeles.
“Through their experiment, which was published in the journal Nature Cities in June, the researchers sought to evenly spread traffic down more corridors as the drivers head to similar destinations, in the hopes of reducing congestion.”
The experiments nudged up vehicle speeds by 2% on average on the congested roads, and shortened travel times by a modest 0.7%. Scientists have long thought that traffic could be eased this way, and this now demonstrates how it could actually work in the real world.
The researchers collected data on weekdays in the 10 cities, which included Seattle, Salt Lake City, Chicago and Atlanta. While they studied a variety of types of roads, from city streets to highways, they found that in cities such as Los Angeles where they targeted highways, their diversions had the biggest effect, with driving speeds in those areas increasing by about 4.6%. Most cities, though, saw a smaller improvement with speeds across them increasing by just 0.5% during peak times. The Google study’s findings confirm what other researchers have long suspected, said Michael Manville, a professor of urban planning at UCLA who was not involved in the study. “This is a pretty nice empirical illustration of a longstanding idea,” he said. If a clogged traffic artery is on the verge of a long delay, he said, rerouting a small number of vehicles away from the bottleneck produces a “not gigantic but noticeable result….”
As with other traffic interventions, there’s a risk that Google’s approach to spreading traffic could backfire. This concept, known as “induced demand,” also applies to efforts like adding a lane to a critical highway, which temporarily relieves congestion, until traffic rises again. “You can do this, but you cannot be too successful, because if you are too successful, then more people will start driving, and essentially you’ll be back to ground zero,” [said Alexandre Bayen, a co-author of the study and director of UC Berkeley’s Center for Information Technology Research in the Interest of Society]. “That’s why when we did the study, we only operated at the margin.” The idea is to make minor improvements, enough to matter in the aggregate but not so significant that people notice and start driving more, canceling those improvements…
The Google Research study also considered the potential environmental benefits of its traffic improvements. The paper highlights the carbon dioxide emission reductions resulting from the slightly smoother traffic flow and shortened travel times, which add up to a savings of about 1,000 metric tons annually. That’s equivalent to the emissions of about 200 vehicles, a drop in the bucket for cities with millions of cars, the majority of which are still gas guzzlers, even in San Francisco and LA. Bayen acknowledges that while emphasizing the cumulative benefit if such efforts were sustained for a longer time. For Manville, the study is a useful proof of concept. “It is a demonstration that if we managed roads better, air would probably be cleaner,” he said.
According to the article, “Google’s study might not just be a one-off, and city officials could decide to make use of the Google research findings, Bayen said, such as by trying to improve traffic congestion on a larger scale…”
Advertisement
“Google declined SFGATE’s requests for comment on whether the company informed city authorities or rerouted drivers about the study,” the article points out.
I have lost count of how many times I have pasted my own old emails into a chatbot just to get it to sound like me. If OpenAI has its way, that annoying step might disappear soon.
The company is testing a new feature called Writing Style inside ChatGPT, and it does something clever. Instead of asking you to explain your tone every single time, it looks at how you already write across your connected apps and mimics it.
How does ChatGPT’s writing style feature work?
Based on the onboarding screen spotted by X user Gael Breton, ChatGPT pulls examples from three buckets: Messaging, Documents, and Email. For example, you can connect Slack for messages, Google Drive or Notion for documents, and Gmail for your email.
Once you connect these apps and hit Use writing style, ChatGPT should reference your actual writing patterns instead of relying on generic prompts. So instead of typing “write this in a casual tone,” you can let your Slack messages do the talking.
Advertisement
How is this different from Claude’s Styles feature?
If this sounds familiar, that is because Anthropic already tried something similar with Claude. Its Styles feature let you upload or paste writing samples, and Claude built a custom style from that. Sadly, Claude deprecated the Styles feature and merged it with Claude Skills, which is not as easy to set up.
Rachit Agarwal / Digital Trends
So ChatGPT is scoring a big win here. Not only is it easier to set up, but instead of manually feeding samples, ChatGPT can pull directly from apps you already use every day. That likely means a much bigger and more accurate sample of your writing, covering everything from quick Slack replies to longer, more formal emails.
Should you be excited?
OpenAI has confirmed it is testing this feature, but there is no word yet on a wider rollout. Still, if you are tired of chatbots sounding a little too robotic, this is a promising step. The idea of an AI that writes emails sounding like you, and not like every other ChatGPT user, is genuinely exciting.
It also fits a bigger trend of AI tools trying to feel less generic and more personal, which honestly cannot come soon enough. For now, we will have to wait and see when, or if, Writing Style reaches everyone.
Two weeks after X Corp’s lawyers demanded it disappear, Nitter is coming back. The open-source project posted a two-line notice on Monday saying it will continue, following legal advice. It did not say what that advice was.
The notice sits on nitter.net, struck through the old cease-and-desist banner. “Following legal advice, the Nitter project will continue,” it reads. “This instance and others will be back up and running shortly.” The maintainer works under the handle zedeus. He promised a fuller announcement within a week.
Richard Speed reported the reversal for The Register. The publication asked X for comment and had received no response.
What TNW could confirm
Some of this is already visible rather than promised. XCancel, the best-known public instance built on Nitter’s code, was serving pages again on Monday. TNW loaded it on Tuesday and it was still up.
Advertisement
The GitHub repository has also come back. Archiving it, which the project did after the letters arrived, freezes a repository and makes it read-only. As of Tuesday it is publicly accessible again and shows no archive notice. The README carries the same update as the website.
The main nitter.net instance has not resumed normal service. So the position is mixed. A project says it is continuing, a repository is live, one large instance is running, and the flagship remains dark.
The letters that started it
X Corp sent cease-and-desist letters on 24 August. They demanded permanent removal of Nitter instances and of the project’s repository. For a while they worked. Instances went dark, the repository froze, and Nitter went offline after seven years.
Nitter is a front end for reading posts on X. It needs no account, runs no JavaScript and carries no advertising. Every request goes through its own backend, so X never sees the reader’s address or browser fingerprint. You cannot post through it. You can only read what is already public.
Advertisement
That last point matters. Nitter republishes no private material and competes for no advertising. It strips the tracking and the account requirement from public posts, and those are the parts X monetises.
The project has never been a stable target either. It reads X through the platform’s unofficial API, which X keeps changing, so each round of restrictions forces a rebuild. Instance operators, not the maintainer, carry most of that cost. That is one reason the number of working public instances has shrunk for years, letters or no letters.
Why the advice might have been encouraging
The project has not said what changed its mind, so this is context rather than explanation. But the legal weather around scraping has shifted, and it has moved in the direction of the scrapers.
In July a US court dismissed Google’s DMCA claims against SerpApi, a company built on scraping Google’s own results. TNW covered the ruling as the moment others can scrape Google. On that reasoning, reading public pages is not automatically infringement or circumvention.
Advertisement
Money is moving on the same assumption. Oxylabs, the Lithuanian web-data firm, raised $130mn from Warburg Pincus in July at a $3.6bn valuation. The premise is that gathering public web data at scale is a legitimate business. A volunteer project reading public posts is a smaller version of the same argument.
None of that is a ruling in Nitter’s favour. A cease-and-desist letter is a demand, not a judgment, and X has not sued. The letters simply asserted what X wanted. Someone appears to have told zedeus those assertions were weaker than they looked.
The asymmetry has not gone away
Whatever the advice said, the underlying imbalance is unchanged. On one side is a company owned by the world’s richest man. On the other is a maintainer with a pseudonym, a Matrix channel and a donations page.
There is a limit to what a letter can undo, though. Nitter is published under the AGPLv3, and licences of that kind are not revocable. Copies already distributed stay licensed, whatever happens to the original repository. That is why demands aimed at a repository rarely remove the software itself, and why one instance stayed reachable while the project decided what to do.
Advertisement
European regulators have started noticing that this asymmetry has consequences. The Cyber Resilience Act pulls unpaid maintainers into compliance obligations built for companies, including a 24-hour disclosure deadline. The recurring problem is the same in both cases. Infrastructure that millions rely on often rests on people with no legal department.
Nitter has been here before, without any lawyers. In January 2024 zedeus declared the project dead after X disabled the guest accounts it relied on. It returned a year later using account sessions. Hosting an instance, he warned at the time, was nowhere near as easy as it used to be.
What is actually resolved
Very little, which is worth saying plainly. The promised announcement has not arrived. X has not commented, has not withdrawn the letters publicly, and retains every option it had two weeks ago. Suing is one of them.
The reason people keep rebuilding this thing is not nostalgia for Twitter. Public institutions, newsrooms, universities and emergency services still post there. Reading them means holding an account with a company many readers would rather avoid entirely. Nitter is the workaround, and tools for reading social platforms on your own terms keep drawing acquirers rather than fading.
Advertisement
What the next week produces is the thing to watch. Either zedeus explains the legal position and the flagship instance returns, or the announcement slips and this becomes another quiet retreat. Two weeks ago the obituary looked safe to write. It was not.
You must be logged in to post a comment Login