Category: Technologies

For those of us who incorporate AI into our daily routines, including myself, personal preferences have already been established. I rely on a premium Claude subscription to assist with my writing tasks (such as brainstorming headlines and adjusting tone), while Gemini serves as my primary tool for creating images and conducting background research, particularly when I need to delve deeply into a subject.

You might prefer Perplexity for searching or ChatGPT for coding, and that is perfectly acceptable since it enhances efficiency and delivers the necessary information quickly. The issue is that your device’s operating system is unaware of your preference and, until recently, did not care before forcing its own AI selection.

In the business sector, this is referred to as partnerships. The majority of Android makers, such as Samsung and OnePlus, have collaborated with Google to embed Gemini directly into their hardware. Conversely, Windows mandates Copilot regardless of user preference.

Shikhar Mehrotra / Techgeeks

Apple, according to reports, is preparing to implement a feature that neither competitor has considered, let alone executed: asking users which AI they prefer and where.

The era of your device dictating your AI choice is nearing its end

The difference becomes immediately apparent when examining the strategies employed by Android and Windows.

According to Bloomberg’s Mark Gurman, iOS 27 will launch a new capability internally dubbed “Extensions.” This feature will enable users to choose their preferred AI model for specific Apple Intelligence functions rather than accepting whatever model Apple decides to install.

Upon the release of iOS 27, users will be able to navigate to Settings, assign their chosen third-party AI model to an Apple Intelligence feature (Writing Tools, Image Playground, etc.), and iOS 27 will handle the rest. Siri will receive similar treatment, allowing you to select the AI model that processes your requests in the background. 

As soon as iOS 27 is available on my iPhone 17, I plan to designate Claude as the AI model for Writing Tools and Gemini for Image Playground. 

Rachit Agarwal / Techgeeks

This goes beyond a simple selection; it is a seamless, system-level preference. Once configured, every time you access an Apple Intelligence tool, your chosen AI model will appear automatically. There is no need to switch applications or copy and paste prompts and results between windows. 

This represents true model portability in action, and honestly, it is a far more logical solution than anything Android or Windows has managed to deliver. 

On Android and Windows, accessing your preferred AI always requires a detour

Consider using AI text tools on a Samsung Galaxy smartphone. Galaxy AI provides numerous options, including Writing Assist, Note Assist, and Call Assist, but all are driven by Google’s Gemini. 

What if I wish to use Claude in Samsung Messages to compose a text? I would need to leave the app, launch Claude’s app or website, paste the text, enter a prompt, copy the result, return to Messages, paste it, and send it. Need a different model to summarize meeting notes in Samsung Notes? The same process applies. 

I know, just reading that is exhausting. Imagine millions of users performing this cumbersome routine daily just to utilize their preferred AI. On Windows, Copilot is hardcoded into Notepad and Paint, leaving everyday users with no other option. 

Nadeem Sarwar / Techgeeks

Motorola deserves recognition for surpassing most Android OEMs by enabling users to access Gemini, Copilot, Perplexity, and Meta’s Llama AI through Moto AI. However, the company has already assigned each AI model to specific use cases. For example, Llama powers Catch Me Up, Copilot handles quick Q&A, and Gemini manages photo analysis. Exciting, perhaps? But I did not choose any of those assignments.   

The platforms offer an AI ecosystem to alleviate digital frustration, but what they have actually constructed is a walled garden. 

Apple isn’t winning the AI race: It’s building the track

Apple’s strategy, conversely, treats your AI preference similarly to how iOS handles default browsers or email clients: as a user setting. In my view, this is genuine democratization. It is the right to choose which AI is truly beneficial to me, rather than whichever ranks highest on a benchmark.

To grasp how Extensions might function in reality, any AI firm could opt in and integrate support via their App Store app, which would then operate as an engine within Apple Intelligence. After installing the app, it will appear as an option in Settings.

From there, you can direct an Apple Intelligence tool to whichever model you trust most for that particular task. While the company’s proprietary models remain available, the ability to outsource a query to third-party models sits as an additional layer. Furthermore, we might see a dedicated App Store section highlighting compatible AI applications.

Andy Boxall / Techgeeks

Here is where it becomes even more intriguing for Apple as a business. Historically, the company has faced criticism for falling behind in the AI race. However, opening its platform of over 2.5 billion active devices, including iPhones, iPads, and MacBooks, to Claude, Gemini, and other qualified providers could transform that weakness into its most profitable strength.

The iPhone maker already collects a standard 30% commission on App Store subscriptions, but if this extends to every Claude Pro or Gemini Advanced subscription processed through iOS, Apple suddenly finds it does not need to win the AI race. It merely needs to own the racetrack.

Keep in mind, none of this has been officially confirmed by Apple. However, we are approaching WWDC 2026, which is when Apple could officially announce the shift from being an AI-first company to an AI-agnostic platform that profits from all of them.

ReMarkable is known for producing some of the best e-ink tablets available. However, prospective buyers have always faced one major hurdle: the steep entry price. The ReMarkable Paper Pro is priced too high, and the Paper Pro Move lacks independent functionality. Recognizing this gap, the company has introduced a new e-ink tablet to resolve the issue.

Introducing the ReMarkable Paper Pure to the ReMarkable lineup, this device could be the ideal choice for those who have considered a ReMarkable product but wanted to avoid premium pricing.

ReMarkable

Priced at $399, the Paper Pure is a 10.3-inch black-and-white e-ink tablet. It includes a Marker pen, with an optional Marker Plus and Sleeve Folio bundle available for $449. At this starting price point, the ReMarkable Paper Pure presents an attractive option, so let’s explore what comes with this new e-ink device.

What’s Included for $399?

The ReMarkable Paper Pure is slim, lightweight, and impressively constructed for its price. Weighing just 360 grams and measuring 6mm thin, it’s easy to carry around. ReMarkable also states that the Paper Pure boasts the longest battery life among its devices, lasting up to three weeks on a single charge with about an hour of daily use.

ReMarkable

According to the manufacturer, the third-generation black-and-white Canvas display is the sharpest and brightest in their lineup. To achieve the target price, ReMarkable made some compromises. The Paper Pure lacks a front light and color display.

However, if you primarily use your tablet for writing and note-taking, you likely won’t miss these features. For instance, I recently bought a Super Nomad without either feature, and I’m very satisfied.

How Is the Writing and Software Experience?

This is the most critical aspect of an e-ink notebook, and the answer is positive. The Paper Pure features the same textured writing surface as the flagship ReMarkable Paper Pro. Digital ink appears in just 21 milliseconds, and page navigation is up to twice as fast as the ReMarkable 2.

ReMarkable

Additionally, you receive all the software features found in the more expensive Paper Pro version. You can take notes, annotate PDFs, and even download web articles for offline reading.

An optional subscription service unlocks extra capabilities. You can sync it with Google Calendar or Microsoft Outlook to automatically generate meeting notes, import files from Google Drive, OneDrive, or Dropbox, and search through handwritten notes. It also includes AI-driven handwriting-to-text conversion.

If you were considering a ReMarkable tablet but were deterred by the cost, the ReMarkable Paper Pure could be the perfect solution.

Google has not yet unveiled a completely autonomous AI agent, but the tech giant is actively building one. According to Business Insider, which examined an internal document, the company is crafting an AI agent known by the codename Remy. Currently, this tool is being piloted by staff members within a restricted, employee-only version of the Gemini application.

Remy is characterized as a “24/7 personal agent for work, school, and daily life” capable of performing actions on your behalf, tracking important matters, proactively managing complex tasks, and adapting to your preferences over time. Google has not provided immediate comments, and no official public release schedule has been established.

OpenClaw’s Viral Success Drives Google’s Entry into the AI Agent Market

OpenClaw

OpenClaw, a free open-source AI agent, recently stunned the tech industry by accumulating over 100,000 GitHub stars in less than a week. This tool can autonomously respond to messages, perform research, manage files, and automate computer tasks without requiring user input.

OpenClaw

Its popularity surged to such heights that Nvidia CEO Jensen Huang hailed it as “definitely the next ChatGPT.” The intense demand for OpenClaw was significant enough to drive up secondhand MacBook prices in China by 15%. OpenAI eventually recruited OpenClaw’s creator.

Remy appears to be Google’s strategy to create a similarly ambitious product but delivered as a refined, seamlessly integrated solution.

The AI Agent Race Intensifies Among Tech Giants

Google’s Remy project underscores that the AI agent sector has become a fierce competition. Anthropic recently introduced Claude Cowork, which can manage PC tasks without the intricate setup OpenClaw demands.

Meta purchased Manus AI and released My Computer, a desktop agent that organizes files, launches applications, and sends emails on your behalf. Meanwhile, Nvidia is developing NemoClaw, an open-source platform that enables businesses to deploy autonomous AI agents across any hardware.

AI Unsplash

While OpenClaw has encountered significant security concerns, with researchers highlighting exposed admin panels, prompt injection vulnerabilities, and plain-text credential storage, Google’s approach is expected to differ. We anticipate a deeply integrated, privacy-focused agent from a trusted platform, which could ultimately dominate the market.

AI Unsplash

Remy is currently undergoing a dogfooding phase, a standard procedure where tech companies test products internally before public release. Google’s I/O event is scheduled for later this month (May 19-20), where it is highly anticipated that the company will unveil its latest AI innovations.

AI agents are expected to be a major focus at this event, and Remy might make its debut public appearance if Google decides to reveal its plans.

Google has not yet released a fully autonomous AI agent, but development is underway. According to Business Insider, which examined an internal report, the tech giant is building an AI agent known by the codename Remy. Currently, staff members are testing this tool within a restricted, employee-only version of the Gemini app.

Remy is characterized as a “round-the-clock personal assistant for work, education, and everyday activities” capable of performing tasks on your behalf, tracking important items, managing complex workflows proactively, and adapting to your preferences as it learns. Google has not provided an official comment at this time, and no release schedule has been confirmed.

OpenClaw’s viral success prompts Google’s market entry

OpenClaw

OpenClaw, a free open-source AI agent, recently captivated the tech industry by accumulating more than 100,000 GitHub stars in less than a week. This tool can autonomously respond to messages, perform research, manage files, and automate computer tasks without user intervention.

Its rapid rise to popularity led Nvidia CEO Jensen Huang to label it “definitely the next ChatGPT.” The high demand for OpenClaw was significant enough to increase secondhand MacBook prices by 15% in China. OpenAI eventually recruited OpenClaw’s creator.

Remy appears to be Google’s strategy to create a comparable product with similar goals, but as a refined and seamlessly integrated solution.

Major tech firms enter the AI agent arena

Google’s Remy initiative confirms that the AI agent sector has become a highly competitive race. Anthropic recently introduced Claude Cowork, a tool that manages PC tasks without the intricate setup OpenClaw demands.

Meta purchased Manus AI and released My Computer, a desktop agent that organizes files, launches apps, and sends emails for users. Additionally, Nvidia is developing NemoClaw, an open-source framework that enables companies to deploy autonomous AI agents on any hardware.

AI Unsplash

While OpenClaw itself has faced serious security scrutiny, with researchers warning of exposed admin panels, prompt injection risks, and credentials stored in plain text. We can expect Google’s version to be a deeply integrated, privacy-conscious agent from a trusted platform, which might be what actually wins this market.

Google’s Remy is currently in a dogfooding phase, which is standard practice at tech companies where employees test products before they reach the public. The company will hold its Google I/O event later this month (May 19-20), where it is widely expected to showcase its next wave of AI products.

Agents are likely to be a centerpiece at this event, and Remy may well make its first public appearance there if Google is ready to show its hand.

Google’s AI-driven Search capabilities have dramatically transformed the way we conduct online research. Rather than wading through endless search results, most users now simply read the AI Overview and move on. Google intends to shift this behavior. The tech giant is introducing five new features to AI Mode and AI Overviews that aim to display more clickable links and provide users with stronger incentives to visit the underlying websites.

Deeper Discovery and Contextual Links

The standout new feature is Further Exploration, a dedicated section that appears at the bottom of AI Overviews, offering curated links to specific articles, case studies, or reports relevant to your query. For instance, if you search for how cities are increasing green spaces, Google might show links to a stream restoration project in Seoul or a report on the design of New York’s High Line park. This feature is designed to encourage users to continue their research rather than closing the tab immediately after reading the summary.

Google

Google is also embedding more links directly within AI responses, positioned next to the relevant text rather than grouped at the bottom. For example, searching for a California bike trip might show a link to a Pacific Coast touring guide next to a bullet point about terrain, and a training blog post next to a bullet point about daily mileage. This approach provides users with a more direct path from the AI answer to the source material behind it.

Google

Google

On desktop, hovering over any inline link will trigger a preview showing the website name and page title, which is aimed at giving users more confidence about visiting the website. Google’s internal testing found that users were more hesitant to follow links when they could not tell where they led, so the preview removes that friction before the click.

Google

Subscriptions and Community Perspectives

AI Mode and AI Overviews will now label links from a user’s active news subscriptions so they stand out in results. Google says early testing showed users were significantly more likely to click those labeled links. For subscribers, it means the publications they already pay for will be easier to find inside AI search results rather than buried below them.

Google

AI responses will also begin surfacing previews of perspectives from public forums, like Reddit, social media, and firsthand sources, with added context like a creator’s handle or community name.

Google

Google

A search about photographing the northern lights, for example, might surface tips from a specific photography forum, with a link to the full discussion thread. Thanks to this, users who want real-world advice rather than a synthesized summary will have a clearer path to the people who have actually been there.

Google

The Bigger Picture

These updates also carry real stakes for publishers. AI Overviews have raised concerns across the media industry about declining referral traffic, and these features are Google’s most direct attempt yet to show that AI search and the open web can coexist. Whether they move the needle on click-through rates will be worth watching.

AI Overview accuracy, however, remains an open question. It has a history of confidently stating wrong information, and the featured image for this story is a reminder of that: it misidentifies today’s date as May 20, 2025. Getting users to click through to publishers may be a step in the right direction, but it’s hard to fully trust a guide that does not always know what day it is.

Microsoft is finally addressing a long-standing complaint from Windows 11 users: File Explorer suffers from sluggishness that a quicker launch time alone cannot resolve.

According to Microsoft’s recent update, the company is implementing a comprehensive performance strategy for the Windows 11 File Explorer. This includes targeted adjustments to startup processes, disk operations, visual rendering delays, and application stability.

These issues impact daily workflows significantly. Tasks like navigating folders, loading thumbnails, opening context menus, and dealing with minor freezes can make the file manager feel unnecessarily heavy and unresponsive.

Why Quicker Launches Aren’t Enough

Microsoft’s previous solution involved preloading File Explorer to keep essential components ready in the background. While this reduces the initial wait time when clicking the icon, it doesn’t address the app’s performance after it opens.

Paulo Vargas / Techgeeks

However, this approach has drawbacks. A Windows Latest test found that preloading consumes approximately 35MB of RAM. While this may seem minor on modern systems, it reinforces the argument that Microsoft is using memory to mask underlying code inefficiencies.

Paulo Vargas / Digital Trends

The real issues occur after the application launches. Large directories may still load slowly, and right-click menus can lag. Microsoft’s new strategy is more robust as it addresses the sequence of startup tasks, eliminates redundant visual processing, reduces disk access, and prevents application freezes.

Inside the File Explorer Update

Tali Roth, Microsoft’s Head of Product for Windows Shell, outlined the File Explorer roadmap as a dual-pronged effort combining startup optimizations with deeper engineering improvements.

When I talk about work we’re doing to make Windows more reliable, performant, and crafted, there’s been a category of comments that I will…politely 😉… summarize as “prove it”. No doubt we’ve got more to do, but excited to share what we’ve delivered so far! https://t.co/oKE94G8b0X

— Tali Roth 🪟 (@TeaAndDates) May 1, 2026

See More

The Windows Shell team is focusing on the most noticeable user experiences. Efforts include optimizing load sequences, stripping out superfluous animations, minimizing disk reads, and stabilizing the app to prevent hangs.

These improvements target areas that preloading cannot fix. The goal is a file manager that opens instantly, populates folders without lag, and keeps thumbnails and menus responsive.

Looking Ahead

Microsoft plans to deploy these File Explorer optimizations gradually over the next few months. The true test will be whether average Windows 11 users perceive a tangible improvement once these changes reach stable releases.

Preliminary feedback is promising. Recent builds have reportedly enhanced navigation smoothness, eliminated the flashing artifact in dark mode, and begun integrating modern UI elements into older File Explorer components.

A specific release date for the general public has not been set. Once the updates arrive, users should monitor everyday interactions—folder loading, thumbnail generation, context menu responsiveness, and overall fluidity. These are the key indicators that Microsoft needs to demonstrate to prove Windows 11’s file manager is genuinely faster.

Xiaomi has frequently faced criticism for drawing inspiration from Apple. However, the roles may soon reverse. The Chinese tech giant has extended its paid battery upgrade initiative in China to include the Xiaomi 13, Xiaomi 13 Pro, and Xiaomi 13 Ultra.

In contrast to standard battery replacements that merely swap depleted cells for identical ones, Xiaomi is providing larger-capacity batteries to enhance the longevity of older devices.

Hold On… A Battery Boost Years After Launch?

Xiaomi 13 Ultra Andy Boxall / Digital Trends

According to information provided by Xiaomi President Lu Weibing on Weibo, the base Xiaomi 13 will see its battery capacity increase from 4,500mAh to 4,850mAh. The Xiaomi 13 Pro will experience a more substantial upgrade, jumping from 4,820mAh to 5,361mAh, while the Xiaomi 13 Ultra will move from 5,000mAh to 5,500mAh. While these increments are smaller than those seen in its latest flagship devices, they represent a significant enhancement for phones that are nearly four years old.

Furthermore, the cost is remarkably low.

The service is priced at 189 Yuan (approximately $27), which covers both parts and labor. Xiaomi cautions that availability might be restricted, indicating this is not yet a globally accessible program. Nevertheless, the concept is compelling. As smartphone manufacturers increasingly promote extended software support and longer device lifecycles, an official higher-capacity battery swap aligns perfectly with this trend. It not only restores but enhances battery performance, much like an OTA update improves software functionality.

Parilov / Shutterstock

Apple Should Take Note

Apple already provides battery replacements for older iPhones, which is commendable. However, Xiaomi’s strategy is more innovative because it transforms a repair into an upgrade. Picture bringing an older iPhone in for service and receiving a certified, higher-capacity battery crafted by Apple, optimized through iOS, and backed by official support. This would make battery replacement more attractive than merely restoring the device to its original battery performance.

The recently released Steam Controller from Valve has already sold out upon launch. However, in a move characteristic of the company, Valve is now empowering enthusiasts by sharing the design blueprints. The company has published the CAD files for the Steam Controller’s outer shell, enabling users to download the exterior specifications and design custom accessories or replacement casings.

The Significance of This Release

New Steam Controller Valve

These files are distributed under a Creative Commons license, making them highly valuable for modders, 3D-printing enthusiasts, and accessory developers. It is important to clarify that these files only cover the controller’s outer shell and do not provide a complete blueprint for rebuilding the Steam Controller from the ground up. Consequently, users cannot 3D print the internal functional components of the gamepad.

New Steam Controller Valve

However, this release does pave the way for creative experimentation and customization. Users can now design alternative grips, custom casings, docking stations, protective cases, puck accessories, and other enhancements based on Valve’s official specifications. This approach highlights an enthusiast-centric design philosophy, making Valve’s controller more attractive to a specialized audience.

Valve’s Distinctive Approach to Hardware

While most companies restrict accessory designs and force third-party developers to rely on measurements, scans, and trial-and-error, Valve is adopting a more transparent strategy by providing official design data from the outset.

Steam Controller ran out faster than we anticipated, and we hate that not everyone who wanted one was able to get it. We’re working on getting more in stock and will have an update on expected timeline soon.

— Valve (@valvesoftware) May 5, 2026

See More

This strategy aligns well with Valve’s broader hardware ethos. Through the Steam Deck, SteamOS, and its long-standing support for PC modding culture, Valve has consistently encouraged user tinkering rather than opposing it. The Steam Controller is just one component of Valve’s expanding hardware lineup, which also features the forthcoming Steam Machine and Steam Frame VR headset.

Currently, the Steam Controller is already sold out, causing frustration among those eager to experience Valve’s innovative PC-friendly gamepad controls. While the release of CAD files does not address the inventory shortage, it does make the accessory feel more open and community-friendly.

The newly launched Steam Controller from Valve has just been released and is already completely sold out. However, in a move that perfectly embodies Valve’s philosophy, the company is now providing the design files to the public. Valve has made the CAD renders for the Steam Controller’s shell available, allowing users to download the exterior design and develop custom accessories or alternative casings.

Why this matters

New Steam Controller Valve

The files are currently available under a Creative Commons license, making them particularly valuable for modders, 3D-printing enthusiasts, and accessory creators. It is important to note that these files only cover the controller’s exterior, not a complete blueprint for rebuilding the Steam Controller from the ground up. Therefore, you cannot 3D print the internal components that make the gamepad functional.

What this does enable is a wide range of experimentation and modification. Users can design alternative grips, custom shells, docks, cases, puck accessories, or other add-ons based on Valve’s official dimensions. This clearly represents an enthusiast-friendly approach, making Valve’s controller more attractive to a niche audience.

How Valve’s controller stands apart

Most companies keep accessories tightly controlled and allow third-party brands to figure things out through measurements, scans, and trial-and-error. Valve is taking a more open approach by providing creators with official design data from the beginning.

Steam Controller ran out faster than we anticipated, and we hate that not everyone who wanted one was able to get it. We’re working on getting more in stock and will have an update on expected timeline soon.

— Valve (@valvesoftware) May 5, 2026

See More

This aligns perfectly with the company’s broader hardware philosophy. Between the Steam Deck, SteamOS, and its long-standing relationship with PC modding culture, Valve has consistently embraced user tinkering rather than resisting it. The Steam Controller is also just one piece of Valve’s larger hardware revival, which includes the upcoming Steam Machine and Steam Frame VR headset.

As of right now, the Steam Controller is already sold out. This early shortage is already frustrating for many who wanted to try Valve’s new take on PC-friendly gamepad controls right away. The CAD file release does not solve the stock problem, though it at least feels less like a closed accessory.

Although Techgeeks Chrome remains the leading web browser, it faces growing competition from the latest generation of AI-driven browsers, such as Perplexity Comet, Dia, and others. To maintain its market position, Google is integrating advanced AI capabilities into Chrome, a move that isn’t inherently negative, yet this particular implementation crosses a line.

Check your file manager for a directory named “OptGuideOnDeviceModel”. If it exists, Chrome has been utilizing your storage as an internal server. Within this folder lies “weights.bin”, a 4 GB file housing Gemini Nano, Google’s local AI model.

Privacy expert Alexander Hanff uncovered and recorded this behavior via macOS file system logs, which monitor all file creation and modification events at the OS level.

On a brand-new Chrome profile with no user interaction, the full 4 GB model installed in under 15 minutes while a tab remained idle.

Was This Action User-Requested?

It wasn’t. Chrome does not seek permission before installing the model; it proceeds automatically. The model downloads silently once Chrome determines your device meets its specifications, long before you interact with any AI functionality.

Google Chrome

If you locate and remove the file, Chrome will re-download it upon the next launch. Hanff observed that “the user’s deletion is treated as a transient state to be corrected, not as a directive to be respected.”

The situation becomes more complex. The most prominent AI feature in Chrome, the “AI Mode” pill in the address bar, does not utilize the local model. Instead, it forwards your queries to Google Gemini servers. The on-device model supports hidden features like “Help me write” in text fields and local scam detection.

What Are the Consequences and How to Disable It?

While this may appear to only impact your device’s storage, Hanff highlighted its broader environmental effect. He calculates that if 500 million devices received this download, the bandwidth alone would generate approximately 30,000 tonnes of CO2 emissions, comparable to 6,500 cars operating for a full year, and this figure only covers delivery, not usage.

Rachit Agarwal / /d

Google must require user confirmation for this download. Currently, you can prevent it by navigating to “chrome://flags”. Search for “Enables optimization guide on device” and disable it. It requires more steps than necessary, but it is effective.