The chief executive officer of HUMAIN, Tareq Amin, has revealed details about the upcoming Horizon Ultra laptop through an unboxing video. The device is set to feature the Qualcomm Snapdragon X2 Elite processor, which will serve as its core computing engine. A HUMAIN logo is also embedded on the right Alt key of the dual-language English-Arabic keyboard located on the left side of the laptop. Amin expressed particular excitement about the operating system of the Horizon Ultra, which is designed with a focus on 'intent' and is intended to support the next generation of agent-based AI. The laptop will be showcased at the LEAP 2026 conference in Riyadh, Saudi Arabia, from August 31 to September 3. The unveiling comes as part of HUMAIN's efforts under the Saudi Arabia Public Investment Fund (PIF) to advance artificial intelligence technologies. The Horizon Ultra represents a significant step in the company's strategy to develop AI-driven hardware and software solutions.
UPDF, developed by Superace Software Technologies, is positioned as a lightweight alternative to Adobe Acrobat, designed to address the limitations of traditional PDF editing in the age of AI. The tool combines a range of functionalities including editing, annotation, conversion, OCR, and an integrated AI assistant, all within a single application. It supports editing text, images, tables, and form fields while preserving the original layout and formatting. This is particularly important for documents such as contracts and financial statements, where maintaining the structure is critical. The tool also offers advanced features such as OCR for scanned documents, which can convert image-based PDFs into searchable and editable text with an accuracy of up to 99%. UPDF supports three output layout options, with the third being especially useful for audit purposes, as it retains the original scan appearance while making the text searchable. Additionally, it allows for converting PDFs into 14 different formats, including Word, Excel, PowerPoint, and image formats, and vice versa. UPDF’s AI assistant, which runs on GPT-5.6, is integrated directly into the editor, enabling users to perform tasks such as generating mind maps, conducting academic research, and creating literature reviews. The AI also includes features like Paper Search and Scholar Research, which help users find and organize academic papers. Version 2.5 introduced ten AI agents, including Copilot, which is designed to identify tools without taking over the user’s actions, ensuring control over sensitive documents. The tool emphasizes data security, stating that documents are not shared with third parties and that content is not used for model training. The free tier allows five uploads, 10 MB per file, 100 pages per file, and 100 questions, with these limits shared across the app and web versions. Paid features include advanced document organization, digital signatures, PDF/A compliance, and enterprise-level auditing. UPDF is priced at one-fifth the cost of Adobe Acrobat, with a 30-day money-back guarantee. UPDF is available on Windows, macOS, iOS, and Android, with support in 12 languages and regular updates. It aims to streamline document workflows by integrating AI capabilities directly into the editing process, addressing the challenges of document automation and ensuring that edits are accurate and maintainable.
Ramp, a corporate expense management platform, has launched its own AI model routing service called Router, aiming to compete with companies like Stripe in the AI inference space. The service allows users and businesses to switch between various large language models via an API, with Ramp having used a similar internal tool for its own AI needs over the past three years. Router is currently available only in the United States and is free to use for the remainder of 2026, though users will still be charged for AI model inference costs. It comes with a $26 credit launch offer, and the company has not disclosed pricing for next year. The service offers access to models from OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai, along with several routing strategies to help customers direct AI requests based on preferences. Ramp also highlighted its opt-out data retention policy, which will store model inputs, outputs, and tool calls for one year by default, with personally identifiable information removed before use. Entering the model routing business provides Ramp with opportunities to tap into the growing AI inference market and integrate its service with existing products like AI token usage monitoring. If successful, Ramp could also build long-term relationships with AI labs and inference providers, potentially expanding its customer base and offering new entry points for its expense management solutions. The company raised $750 million at a $44 billion valuation in June. Ramp’s move into AI model routing follows recent developments in the industry, including Stripe’s reported $7 billion acquisition of OpenRouter.
On August 21, RayNeo launched its new product line, the RayNeo iO, positioning it as the first 'Human Augmentation AI Glasses'. The device aims to integrate AI into human vision, perception, and action, becoming a personal intelligent enhancement companion. IT Home has already received a pre-release unit and provided an unboxing review. The packaging of the RayNeo iO AI Glasses is described as high-end and elegant, with a minimalist design. The box is silver-gray in color, featuring a metallic 'RayNeo' logo and a rectangular window that extends to the side, displaying the product name. The overall design is restrained. Inside the box, the glasses are placed in a dedicated leather case, with accessories including a user manual, warranty card, cleaning cloth, screwdriver, silicone lens tips, charging clip, and nose pad set. Each accessory comes in its own small box, enhancing the product's premium feel. When handling the RayNeo iO AI Glasses, one might initially be confused—could this really be a smart glasses? Without the functional modules at the end of the temples, it resembles a regular fashion accessory. The glasses achieve a slim and elegant look. The size of the light engine determines the front frame's appearance. A larger light engine makes the temples bulkier, making it obvious that it is a smart glasses. The purity of the waveguide affects the wearer's comfort. The RayNeo iO uses a new Nano light engine with a volume of 0.085 cubic centimeters, roughly the size of a red bean. The new Blue Lake waveguide is less than 0.6 mm thick, weighing only 3 grams, with an OC area transparency of 93% and non-display area transparency exceeding 98%. Whether on or off, the glasses look almost like regular optical lenses, eliminating the bulkiness and odd appearance of previous smart glasses. The front frame of the RayNeo iO AI Glasses is designed with a classic crown shape, suitable for various face types. According to the official description, the frame is made of AZ91D magnesium alloy, with an average edge width of 2.25 mm. The temples are tilted 5 degrees to fit different facial contours, with two sets of replaceable nose pads and special silicone temple tips. The RayNeo iO is available in two color options: 'Eclipse' and 'Sand Dune'. The unit IT Home received is the 'Eclipse' version, a matte black with a sand-like texture. On the right temple, there is a Smart Crown knob, which serves as the primary control interface for the glasses, alongside head gesture interaction, allowing users to quickly confirm notifications in public settings by nodding or shaking their head. The middle section of the temple is made of aerospace-grade titanium alloy, with β titanium 15333 as the main body and TC4 as the outer cover. It is very thin, with the thinnest part measuring 1.15 mm. Other components are integrated into the ends of the temples, achieving a balanced weight distribution without affecting wear. The official weight of the RayNeo iO AI Glasses is 34 grams, making it very lightweight. Combined with the balanced weight distribution design, the glasses do not cause noticeable pressure on the nose or ears during extended wear. The replaceable nose pads and silicone temple tips also contribute to a comfortable wearing experience. In addition to the glasses, IT Home received a charging case for the RayNeo iO, featuring a 2000 mAh battery that can fully charge the glasses five times. The glasses can be automatically charged by placing them in the case. The case is designed with a combination of velvet leather and semi-transparent plastic, with a USB Type-C port on the outside for charging. The overall design and materials convey a sense of high-end quality. Overall, the RayNeo iO AI Glasses feature a minimalist, business-oriented design with sharp lines. The main highlights are its lightness and balance, making it both aesthetically pleasing and comfortable for daily wear.
MiniMax has launched a new multimodal creation platform called MiniMax Design, aimed at enhancing the productivity of its native multimodal video model, H3. The platform is designed to integrate seamlessly with H3's capabilities, including multimodal understanding, video generation, editing, and reference control, to streamline the content creation process. MiniMax Design functions as a multimodal creation harness, enabling users to input creative requests and allowing the AI agent to interpret the task, break it down into actionable steps, and leverage the appropriate models and skills to handle content generation, editing, and final delivery. The platform addresses challenges in task planning, material organization, and multi-step collaboration, allowing the model to participate in the entire content creation process from ideation to final delivery. As part of MiniMax's ecosystem around H3, MiniMax Design incorporates insights into model capabilities, input methods, and usage patterns into its agent and skill sets. When presented with a specific task, MiniMax Design identifies the necessary textual, visual, video, and audio elements, recognizes content that needs to be preserved or referenced, and translates the creative request into suitable materials and instructions for H3. It also enhances pre-production planning, such as storyboard creation, and post-production capabilities, including editing and final assembly of the content. The release of MiniMax Design marks a significant step in integrating advanced AI models into commercial content production workflows, expanding the practical applications of H3 in video creation and editing.
Meta has released a native Mac application for its Meta AI tool, now available in a beta version. The app is designed for businesses and content creators and integrates with Facebook and Instagram. It includes built-in analytics to address account performance queries and supports connections to Google Workspace for document access, though this feature requires a professional Facebook or Instagram account. Users can share their screen windows to allow Meta AI to provide suggestions based on the content being worked on, and the app also features dictation capabilities usable across all Mac applications. Additionally, Meta is expanding AI functionalities for enterprise users across its platforms, enabling AI assistants to handle repetitive tasks, set reminders, and generate presentations, documents, and spreadsheets. Privacy concerns are highlighted, as Meta’s privacy policy states that interactions with Meta AI will be used to train its models. Meta AI is free to use, but a subscription plan called Meta One is available to increase usage limits.
Meta has launched a new Mac application specifically designed for its AI chatbot. The app allows users to share their screen with the AI, enabling it to offer suggestions, answer questions, or generate content based on the information displayed. The application also includes support for dictation across all compatible apps. The release marks an effort by Meta to enhance its AI chatbot as a productivity-focused tool, aiming to better compete with other AI assistants that already have desktop versions. Users can share their windows with Google's Gemini AI app, while OpenAI's ChatGPT and Anthropic's Claude apps offer more advanced features by allowing the chatbots to interact with the content on the screen. The Verge reported on the development.
Google has integrated an artificial intelligence assistant into its new Pixel 11 smartphone, which is capable of performing tasks such as ordering groceries, booking restaurant tables, and taking photos on behalf of the user. However, the functionality of the AI assistant has been met with skepticism, as it appears to lack the practical utility that users might expect. The AI's ability to assist with everyday tasks is seen as a promising feature, but its effectiveness in real-world scenarios remains questionable. Critics are questioning whether such capabilities are truly desired by the average consumer. The AI assistant is part of a broader trend in the tech industry, where companies are increasingly focusing on integrating artificial intelligence into consumer products. Despite the potential benefits, the Pixel 11's AI assistant has not yet demonstrated the level of sophistication or reliability that would make it a compelling addition to the user experience.
Calendly, the company known for its scheduling software, has announced the launch of an AI-powered meeting note-taker aimed at enhancing productivity in the workplace. The new tool joins meetings, records audio and video, and transcribes conversations to generate summaries, action items, and draft follow-up emails. It also includes a Granola-like feature that uses system audio for transcription. Calendly plans to release an AI assistant named Callie, which will leverage the company’s existing meeting and scheduling tools to manage appointments, check availability, and provide context from prior meetings. Calendly CEO Tope Awotona emphasized the need to automate workflows for users in outward-facing roles such as sales and marketing, who often have a high volume of meetings. The note-taker will inform participants that the meeting is being recorded, sending a message in the chat to notify everyone and allowing users to request the app to leave the conversation. Calendly’s note-taker will also inform participants before meetings start through integration with its scheduling system. The competition in the note-taking space is intense, with companies like Granola, Fireflies, Read AI, Otter, and Fathom already offering similar services. Productivity platforms such as Notion, ClickUp, and Wispr have also added note-taking features to their suites. Awotona highlighted the opportunity in automating tasks after meetings, noting that while many note-takers excel at recording and transcribing, the real value lies in post-meeting efficiency. Some companies in the space, including Otter and Granola, have faced privacy allegations, prompting Calendly to prioritize transparency in its note-taking solution.
Replit, uma plataforma de desenvolvimento de software em nuvem, anunciou a introdução de uma nova modalidade gratuita, denominada Free Mode, que utiliza a tecnologia GPT-5.6 Luna. Esta atualização permite que qualquer pessoa transforme ideias em software funcional, sem preocupações com custos de tokens. A funcionalidade é parte de uma estratégia da empresa para democratizar o acesso à criação de software, tornando o processo mais acessível e eficiente. A nova versão do Free Mode é integrada ao ambiente de desenvolvimento da Replit, oferecendo ferramentas de codificação e assistência linguística baseadas na inteligência artificial. A iniciativa visa reduzir barreiras financeiras e técnicas para desenvolvedores iniciantes e pequenas equipes. A Replit destaca que a implementação da GPT-5.6 Luna no Free Mode representa um avanço significativo na capacidade da plataforma de suportar a criação de software complexo com recursos limitados. A empresa não forneceu detalhes sobre os custos associados ao uso da tecnologia GPT-5.6 Luna no Free Mode, mas enfatizou que a experiência será semelhante à versão paga da plataforma, com acesso a recursos avançados de programação e assistência de IA. A introdução do Free Mode está alinhada com o objetivo da Replit de expandir o alcance da programação e da inovação tecnológica, especialmente em contextos de baixo custo e alta acessibilidade.
Amazon has announced that its AI-powered Alexa+ will be made available for free on all compatible Fire TV devices in the U.S., regardless of whether users have an Amazon Prime subscription. The update includes features such as conversational search, smart home controls, and AI-driven recommendations. Previously, Alexa+ was a paid service costing $19.99 per month for non-Prime members and was initially available only on newer Fire TV devices launched last fall. Now, all compatible devices, including current-generation Fire TV Sticks, Fire TV Cube, Amazon Ember smart TVs, and select models from Hisense and Panasonic, will be automatically upgraded to Alexa+. This move aligns with industry efforts to integrate AI into more consumer electronics, similar to Google’s rollout of Gemini on Google TV and Roku’s AI voice assistant upgrade. Amazon claims that users of Alexa+ are engaging more with the AI, with nearly twice as many conversations compared to the original Alexa. The AI can also help manage smart home devices, such as displaying Ring camera feeds on the TV. The decision follows a broader trend of making AI services non-optional on consumer devices.
Xiaomi unveiled the second generation of its smart camera featuring video calling capabilities, with crowdfunding orders beginning on August 26 at a price of 429 yuan. The new model features a 3.97-inch color display with 480-by-800-pixel resolution and enlarged buttons designed for easier use by elderly and children. The device enables bidirectional video calls between the camera and smartphones, or between cameras, offering three activation modes: gesture recognition, single tap, or linked calls. In audio performance, the model received significant improvements with dual integrated microphones and a larger-diameter speaker, ensuring noise reduction and clearer sound reproduction. The camera now reaches 8 megapixels with 4K recording capability (3840-by-2160 pixels), while the internal processor with 1.5 trillion operations per second delivers 50% more performance than the previous generation, enabling more accurate and complex scene detection. The new electronic pan-tilt provides 360-degree horizontal viewing and manual tilt up to 30 degrees. The product integrates with the Xiaomi Xiaoai Link ecosystem, enabling sharing with multiple family members and viewing across smartphones, tablets, and televisions.
Code discovered in macOS Tahoe 26.7 files points to an upcoming Apple product called Home Hub (internal code J490), expected to launch later this year. The device, powered by Siri, will feature a Widget Gallery allowing users to choose which information appears on the home screen. The Widget Gallery will function similarly to what is already available on iPhone, iPad, and Mac. Widgets installed on iPhone may be automatically recommended for the Home Hub, and evidence suggests third-party apps will also receive widget support. When an iPhone widget appears on the device, its data will remain synchronized. For privacy reasons, widgets will only display their content when the primary associated device is within the home. All users will be able to see which widgets have been installed, but not their data. The Home Hub will also feature Smart Stack, a capability that automatically switches between different widgets based on time of day, similar to what is available on Apple Watch. There will also be a Face Gallery for selecting different clock styles. The device will have a square display of approximately 7 inches and can be installed either on a stand or mounted on a wall. It will integrate with HomeKit scenes and automations. It will not feature an independent App Store but will work with Apple apps such as Safari, Apple News, Apple Music, Calendar, Notes, Photos, and Home.
Chinese startup Shangjian Technology unveiled its new loomos L1 smart glasses, designed with a focus on "active memory" and extended comfort. The device combines an aerospace-grade titanium chassis with a natural cellulose acetate frame, weighing just 43 grams. Unlike other AI glasses inspired by sunglasses design, the company engineered it for conventional eyeglass wearers, supporting up to 16 hours of continuous use. The loomos L1 features a 12-megapixel camera with a 111-degree ultra-wide angle. Its primary functions include audio recording to automatically generate an AI diary that merges real-world recordings with smartphone notifications and creates intelligent summaries. The device also offers AI Live Photo, which combines photography with memory features. Technically, it uses a Qualcomm Snapdragon W5100 processor at 4 nanometers paired with a Hengxuan BES2700 Bluetooth chip, running Android. The removable battery has 258mAh capacity, delivering up to 40 hours of runtime when used with the charging case. With active memory enabled, usage drops to approximately 12 hours. Software integration includes Feishu and Tencent WorkBuddy, enabling voice-controlled message management, scheduling, and task delegation. Pricing starts at 2699 yuan for the version with 1.67 refraction lenses from Essilor and reaches 2999 yuan for Zeiss optics. The launch took place in Wuhan, with the company's flagship store opening on August 19.
Xiaomi's CFO revealed in an earnings call that the company will soon launch its first personal desktop application for MiMo, the company's AI model, designed to help users complete integrated work and daily life tasks. The company is developing a next-generation version of the model, with launch expected shortly. On the commercial front, Xiaomi disclosed that MiMo's token plan already contributes revenue with accelerating growth. However, the AI division remains in heavy investment phase, with strategic focus on integrating technology across its device, connectivity, and smart home ecosystems rather than prioritizing immediate monetization. MiMo V2.5 ranked first in monthly and weekly call volume on the OpenRouter platform in July, attributed to Xiaomi's systematic advantages in inference efficiency and cost control. In June, Xiaomi launched MiMo Code V0.1.0, an experimental programming assistant running in terminal environments, featuring the V2.5 multimodal model and supporting integrations with popular third-party AI platforms.
Cursor, the AI development startup now owned by SpaceX, unveiled Origin this week—a new code-hosting platform aimed at replicating GitHub's core functions. The platform supports collaborative coding, repository management, and pull requests, with additional agent-native features planned for the future. Unlike some competitors, Origin is designed to work alongside GitHub rather than replace it entirely; developers can sync repositories between the two platforms seamlessly. The launch comes amid mounting frustration with GitHub's reliability. The platform suffered a major worldwide outage lasting over six hours on the same day Origin debuted, affecting nearly 20% of requests globally. This incident reflects a broader pattern: GitHub has experienced 257 outages over the past year, prompting numerous high-profile developers to consider alternatives. Despite these setbacks, GitHub remains dominant, with approximately 180 million users as of last October, presenting a significant challenge for Cursor's ambitions to gain market share.
The AR brand RayNeo has officially announced its new "iO" smart glasses series, with the first model set to be unveiled on August 21. The name "iO" references the input and output concept from computing, reflecting the company's vision of a new personal intelligent terminal. The RayNeo iO glasses are positioned as "human-enhanced AI glasses," designed to provide information while helping users see greater detail, retain what they learn better, and express themselves more clearly. The project redesigned the internal structure, optical system, and frame form, embedding the display module into a classic crown-style frame design and utilizing a new proprietary waveguide optical technology that renders the screen nearly invisible, allowing information to appear naturally in the user's field of view. The device combines thin titanium arms with a retro aesthetic, creating a lighter appearance closer to conventional eyeglasses. The software emphasizes AI capabilities across knowledge, expression, and memory, offering assistance in communication and content recording scenarios. Pre-orders are now open, with detailed specifications, features, and pricing to be revealed during the official launch event.
OpenAI has introduced a specialized ChatGPT mode designed specifically for adolescents aged 13 to 17. The feature automatically activates for users who identify as teenagers or are estimated to fall within this age range. The new experience combines previously existing safety measures with enhanced protections, incorporating healthy-use features and additional parental controls to support responsible AI usage among young people. The launch reflects growing public concerns about artificial intelligence's impact on youth, particularly amid similar efforts by other platforms to implement age verification and teen-specific safeguards. According to OpenAI, the platform aims to help teenagers learn effectively, develop critical thinking skills, and build confidence in using AI tools.
A video discovered in the macOS Tahoe 26.7 Release Candidate provides the first detailed look at Apple's upcoming AirPods with integrated cameras. The footage shows someone using the new earbuds to capture visual data from everyday objects, such as book covers, leveraging a feature called Visual Intelligence. According to a Siri voiceover in the clip, users will be able to save items they encounter simply by asking the voice assistant. Industry analyst Mark Gurman had previously confirmed that these camera-equipped AirPods are intended to process low-resolution visual input and function as wearable visual sensors.
Alibaba introduced HappyShrimp 1.0, an artificial intelligence model for music generation that enables users to transform emotions, stories, or memories into complete compositions. The system was designed with the goal of lowering barriers to music creation, functioning through natural language descriptions instead of requiring technical knowledge in music theory or composition. The model generates music in an integrated manner, understanding music as a language with syntax, semantics, and context. When processing requirements for descriptions, lyrics, genre, emotional tone, era style, and voice type, the system works in a unified way rather than separating each element. According to developers, HappyShrimp can understand both technical expressions from the music industry, such as "Lo-fi R&B", as well as emotional contexts, scenarios, and narratives described by the user, overcoming common limitations in AI music tools that struggle to comprehend cultural nuances and aesthetics in natural language.
