Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More In a new case study, Hugging Face researchers have demonstrated how small language models (SLMs) can be configured to outperform much larger models. Their findings show that a Llama 3 model with 3B parameters can outperform the 70B version of the model in complex math problems. Hugging Face has fully documented the entire process and provides a roadmap for enterprises that want to create their own customized reasoning models. Image source: Hugging Face Scaling test-time compute The work is inspired by OpenAI o1, which uses extra “thinking” to solve complex math, coding and reasoning problems. The key idea behind models like o1 is to scale “test-time compute,” which effectively means using more compute cycles during inference to test and verify different responses and reasoning paths before producing the final answer. Scaling test-time compute is especially useful when there is not enough memory to run a large model.  Since o1 is a private model and OpenAI has remained tight-lipped about its internal workings, researchers have been speculating about how it works and trying to reverse engineer the process. There are already several open alternatives to o1. Hugging Face work is based on a DeepMind study released in August, which investigates the tradeoffs between inference-time and pre-training compute. The study provides comprehensive guidelines on how to balance training and inference compute to get the best results for a fixed budget. In addition to using extra inference-time compute, the success of the technique hinges on two key components: a reward model that evaluates the SLM’s answers, and a search algorithm that optimizes the path it takes to refine its answers. Image source: Hugging Face Different reasoning algorithms The simplest way to use test-time scaling is “majority voting,” in which the same prompt is sent to the model multiple times and the highest-voted is chosen. In simple problems, majority voting can prove useful, but its gains quickly plateau on complex reasoning problems or tasks where errors are consistent across generations. A more advanced reasoning method is “Best-of-N.” In this technique, the SLM generates multiple answers, but instead of majority voting, a reward model is used to evaluate the answers and choose the best one. “Weighted Best-of-N,” a more nuanced version of this method, factors in consistency to choose answers that are both confident and occur more frequently than others. The researchers used a “process reward model” (PRM) that scores the SLM’s response not only on the final answer but also on the multiple stages it goes through to reach it. Their experiments showed that Weighted Best-of-N and PRMs brought the Llama-3.2 1B near the level of Llama-3.2 8B on the difficult MATH-500 benchmark. Image source: Hugging Face Adding search To further improve the model’s performance, the researchers added search algorithms to the model’s reasoning process. Instead of generating the answer in a single pass, they used “beam search,” an algorithm that guides the model’s answer process step by step. At each step, the SLM generates multiple partial answers. The search algorithm uses the reward model to evaluate the answers and chooses a subset that is worth further exploring. The process is repeated until the model exhausts its inference budget or reaches the correct answer. This way, the inference budget can be narrowed to focus on the most promising answers. The researchers found that while beam search improves the model’s performance on complex problems, it tends to underperform other techniques on simple problems. To address this challenge, they added two more elements to their inference strategy. First was Diverse Verifier Tree Search (DVTS), a variant of beam search that ensures that the SLM doesn’t get stuck in false reasoning paths and diversifies its response branches. Secondly, they developed a “compute-optimal scaling strategy,” as suggested in the DeepMind paper, which dynamically chooses the best test-time scaling strategy based on the difficulty of the input problem.  The combination of these techniques enabled Llama-3.2 1B to punch above its weight and outperform the 8B model by a significant margin. They also found that the strategy was scalable, and when applied to Llama-3.2 3B, they were able to outperform the much larger 70B model. Not a perfect solution yet Scaling test-time compute changes the dynamics of model costs. Enterprises now have the ability to choose where to allocate their compute resources. For example, if you are short on memory or can tolerate slower response times, you can use a small model and spend more inference-time cycles to generate more accurate answers. However, test-time scaling also has its limitations. For example, in the experiments carried out by Hugging Face, researchers used a specially trained Llama-3.1-8B model as the PRM, which requires running two models in parallel (even if it is much more resource-efficient than the 70B model). The researchers acknowledge that the holy grail of test-time scaling is to have “self-verification,” where the original model verifies its own answer as opposed to relying on an external verifier. This is an open area of research. The test-time scaling technique presented in this study is also limited to problems where the answer can be clearly evaluated, such as coding and math. Creating reward models and verifiers for subjective tasks such as creative writing and product design requires further research. But what is clear is that test-time scaling has generated a lot of interest and activity and we can expect more tools and techniques to emerge in the coming months. Enterprises will be wise to keep an eye on how the landscape develops. source

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More OpenAI has expanded the number of applications its desktop apps can work with, including allowing Advanced Voice Mode to work with other apps, and is moving closer to ChatGPT using computers.  The desktop app introduced integrations in November with an initial four applications. During Day 11 of its “12 Days of OpenAI” event, OpenAI announced several new integrated development environments (IDEs), terminals and text apps it will support. ChatGPT now supports BBEdit, MatLab, Nova, Script Editor, TextMate as IDEs, VS Code for VSCode Insiders, VSCodium, Cursor, WindSurf, the JetBrains family of IDEs of Android Studio, AppCode, CLion, DataGrip, GoLand, IntelliJ IDEA, PHPStorm, PyCharm, RubyMine, RustRover and WebStorm. It also added the Warp and Prompt terminal apps as integration. These applications join VS Code, Xcode, Terminal, iTerm 2 and TextEdit as integrated apps.  But coding applications won’t be the only applications that ChatGPT desktop apps can access. OpenAI also added Apple Notes, Notion and Quip to its integrations. Advanced Voice Mode can work with these applications, considering the context of projects in the integrations.  OpenAI emphasized that users must give ChatGPT permission to access these applications.  Letting ChatGPT use your computer for you App integrations with AI chatbots are, of course, nothing new. In October, GitHub Copilot added coding platform integrations. Connecting applications to ChatGPT or Copilot brings context from those platforms into the chat experience. Developers can prompt ChatGPT for some coding help for a project they have on VS Code, and the chatbot understands what they’ve been working on. Kevin Weil, chief product officer at OpenAI, said during a live stream that improving the desktop app will help the company get closer to a more agentic user experience for ChatGPT.  “We’ve been putting a lot of effort into our desktop apps,” said Weil. “As our models get increasingly powerful, ChatGPT will more and more become agentic. That means we’ll go beyond just questions and answers; ChatGPT will begin doing things for you.” He added that the desktop apps “are a big part” of that transformation.  “Being a desktop app you can do much more than you can than just a browser tab,” he said. “That includes things like, with your permission, being able to see what’s on your screen and being able to automate a lot of the work you’re doing on your desktop. We’ll have a lot more to say on that as we go into 2025.” If OpenAI lets ChatGPT see more of your computer, ChatGPT will get closer to Anthropic’s Claude Computer Use feature, which allows Claude to click around a person’s computer, navigate screens and even type text.  OpenAI already announced a fairly similar feature for the mobile version of ChatGPT, although the chatbot cannot yet access computers or phones the same way. Users can share their screens with the chatbot so it can “see” what they’re reading or looking at. Microsoft and Google also developed comparable features with Copilot Vision and Project Astra.  How to access  On MacOS, users who want to open ChatGPT while using other applications can use “option + space” to pull up ChatGPT and choose the application they need through a button on the chat screen. Another shortcut, “option + shift + 1,” brings up the topmost application used.  From that window, users can also access Advanced Voice Mode in the same way. Voice mode automatically detects context from the application.  Integrations are available to ChatGPT Plus, Pro, Team, Enterprise and Edu users. However, Enterprise and Edu subscribers must ask their IT administrators to turn on the feature.  source

Best for Windows: X-Mouse Best for Macs: Mac Mouse Fix Best for Linux: Input Remapper Best for scroll wheel customization: WizMouse Best for recording macros: Macro Recorder Computer mice have been used for decades and are easily taken for granted. Even so, it is possible to make your mouse a bit more appealing by customizing its behavior. Several utilities exist that allow you to reassign mouse buttons or scroll wheel’s behavior. There are even utilities that can record macros of your mouse movements. This article lists five such utilities. 1. X-Mouse Image: X-Mouse X-Mouse is arguably the most popular mouse mapping software for Windows devices. It allows users to assign specific functions and macros to different buttons on their mouse. For example, if you hover over a YouTube video, the scroll wheel on your mouse will automatically adjust the volume rather than scroll up or down. You can create different profiles for different applications and simulate complex keyboard inputs with a single click. X-Mouse is ideal for gamers and power users looking to increase their productivity. Features Customizable button assignments, profile switching per application Price Free SEE: How to Connect an Apple Wireless Keyboard to Windows 10 and Windows 11 2. Mac Mouse Fix Image: Mac Mouse Fix Mac Mouse Fix is perfect for anyone frustrated by the default macOS mouse settings. It brings all the features of an Apple trackpad, such as Quick Look and Zoom, to a third-party mouse. Mac Mouse Fix also addresses common issues like pointer acceleration and scroll smoothing and allows for customization of mouse actions that are activated by clicking, dragging, and scrolling. Finally, it is 100% open-source to provide peace of mind. Features Scroll smoothing, customizable actions Price $2.99, after a 30-day free trial 3. Input Remapper Image: Linux Uprising Blog If you’re a Linux user, the best tool for customizing your mouse is Input Remapper. The creator, sezanzeb, has made it available for free on GitHub. It has a simple Graphical User Interface that makes it easy to assign mouse actions to a different button and set up timed macros. On top of computer mice, the program can be used to map keyboards, D-Pads, and joysticks. Features User-friendly interface, timed macros, supports gamepads Price Free Must-read Apple coverage 4. WizMouse Image: Make Use Of WizMouse is a free utility that can be used to customize mouse behavior. The utility enables the mouse wheel for applications that do not natively support its use. There is also a handy “reverse” function for use when the mouse scrolling is backward. WizMouse can be enabled or disabled from the system tray. Features Scrolling in a non-active window, scrolling in applications that don’t support a mouse wheel. Price Free 5. Macro Recorder Image: Macro Recorder Macro Recorder can be used to automate repetitive tasks you complete using your mouse. The tool records mouse movements and clicks, as well as keyboard inputs, and the resulting macros can be played back to save the user time and effort. The program also includes options for fine-tuning recorded actions to fit specific needs, such as by Features Records macros to automate mouse-led tasks. Price Starts from $111 for the standard version. What is mouse button remapping, and why is it useful?​ Mouse button remapping assigns the default functions of mouse buttons to different actions. This can be done using software that allows users to customize each button’s behavior, such as executing a keyboard shortcut or launching an application. This is useful because, depending on how you use your device, it can enhance productivity and decrease repetitive strain. For example, gamers can assign complex combos to a single mouse button, while professionals can streamline workflows by assigning custom shortcuts. How can I remap my mouse buttons on Windows? There is a limited amount of remapping you can do on Windows 11 without downloading third-party software. To do so, go to Settings → Bluetooth & devices → Mouse. Under the “Primary mouse button” menu, you can switch the primary and secondary, or left and right, mouse buttons. Scroll-related settings can be found under “Scrolling,” and you can choose for your pointer to display a trail across the screen when it is moved by selecting the “Display pointer trails” checkbox under “Pointer Options.” If you want to control the mouse with your numeric keyboard, navigate to Accessibility in Settings, under “Interaction” select “Mouse,” and then turn on the “Mouse keys” switch. Brien Posey contributed to this article. source

The cybersecurity landscape in 2024 was marked by unprecedented challenges, significant breaches, and evolving regulatory requirements that fundamentally reshaped how organizations approach data protection. From record-breaking incidents to stringent new legislation, the year provided crucial insights into cybersecurity. It highlighted critical priorities for strengthening organizational defenses in an increasingly complex digital ecosystem. The escalating sophistication of cyber threats and the expanding attack surface created by digital transformation initiatives posed unprecedented challenges for organizations across all sectors. Record-breaking breaches define the year 2024 witnessed several devastating cybersecurity incidents that underscored the growing sophistication of threats: The year began with the ongoing effects of the MOVEit supply chain breach, which impacted over 2,600 organizations and exposed 77 million records. This incident highlighted the cascading effects of supply chain vulnerabilities in an interconnected digital world and sparked a renewed focus on third-party risk management across industries. The National Public Data breach was particularly severe, compromising 2.9 billion records and affecting 1.3 million individuals. The unprecedented scale of this breach sent shockwaves through the cybersecurity community and prompted many organizations to reassess their data protection strategies. The healthcare sector faced a major crisis with the Change Healthcare breach, which impacted 110 million Americans, underscoring the critical importance of robust data protection measures in handling sensitive medical information. The breach exposed vulnerabilities in healthcare systems and led to nationwide disruptions in patient care and medical billing processes. AT&T experienced cyber incidents exposing 110 million customer records, resulting in an estimated $19.69 billion in financial losses. These incidents demonstrated the severe consequences of inadequate cybersecurity practices and the long-lasting effects on customer trust and corporate financial health. The breaches led to extensive regulatory scrutiny and prompted calls for enhanced telecommunications sector security standards. The financial toll of data breaches continued to rise dramatically, with the global average cost reaching $4.88 million — a 10% increase from 2023. Moreover, 60% of organizations reported spending over $2 million annually on data breach litigation costs alone. These escalating costs can be attributed to various factors, including the increasing sophistication of cyber threats, the expanding attack surface created by remote work arrangements, and growing regulatory consequences. Organizations also faced significant indirect costs, including reputational damage, lost business opportunities, and decreased customer confidence. SEE: US Sanctions Chinese Cybersecurity Firm for 2020 Ransomware Attack Tool sprawl and third-party risks emerge as critical concerns The year also revealed significant vulnerabilities created by complex technology environments and third-party relationships. Organizations using seven or more communication tools experienced 3.55 times more breaches than average, emphasizing the dangers of tool sprawl. While enabling greater collaboration and productivity, this proliferation of communication platforms created new vulnerabilities that cybersecurity professionals struggled to address. The challenge of maintaining consistent security controls across multiple platforms emerged as a critical priority for security teams. The risk landscape was further complicated by organizations’ increasing reliance on external partners, with 66% of companies exchanging sensitive content with over 1,000 third parties. This dependency contributed to a 68% increase in software supply chain attacks targeting file transfer systems. The challenges of tracking and controlling external content sharing highlighted the need for comprehensive data protection strategies that extend beyond organizational boundaries. Many organizations implemented new vendor risk management programs and enhanced their third-party security assessment processes in response to these challenges. Regulatory landscape grows more complex 2024 saw substantial regulatory developments that transformed the data privacy landscape. Implementing the NIS 2 Directive introduced personal liability for cybersecurity compliance violations in the European Union, raising the stakes for executives and boards. This shift toward individual accountability emphasized the need for top-down commitment to data protection and integrating cybersecurity considerations into overall business strategy. Organizations scrambled to update their governance structures and compliance frameworks to address these new requirements. In the U.S., several states passed comprehensive privacy laws, creating a complex patchwork of requirements for organizations to navigate. This regulatory expansion led to significant financial consequences, with GDPR and HIPAA enforcement resulting in fines totaling $5.6 billion and $5.3 billion, respectively. The complex regulatory environment particularly impacted North American organizations, with 63% citing state privacy laws as a top concern, highlighting the need for harmonized and consistent data protection regulations. Many organizations have invested heavily in compliance management systems and privacy program enhancements to address these evolving requirements. SEE: Patch Tuesday: Microsoft Patches One Actively Exploited Vulnerability, Among Others Must-read security coverage Emerging threats and industry-specific challenges The rise of artificial intelligence and machine learning introduced new security challenges, with 50% of North American organizations identifying AI/GenAI data exposure as a primary concern. While offering tremendous innovation potential, these emerging technologies require organizations to develop new strategies for managing unique security challenges. The rapid adoption of AI tools raised concerns about data privacy, model security, and the potential for AI-powered cyberattacks. Cloud security emerged as another critical challenge, with cloud environment intrusions increasing by 75% year-over-year and 33% of breaches tied to misconfigurations. The case for single-tenant versus multi-tenant cloud hosting gained significant attention as organizations sought more secure cloud deployment options. Security teams focused on implementing enhanced cloud security posture management tools and improving their cloud security architectures. The threat landscape evolved significantly, with malware-free attacks comprising 75% of detected incidents and ransomware payments rising by 500% to reach an average of $2 million. Employing an AI-enabled algorithm, we scored different industry sectors from 2018 through 2024, with hospitality, retail, and manufacturing receiving the top risk scores for the first half of 2024. The education and research sector experienced the highest weekly attacks at 3,086 — a 37% year-over-year increase. This highlighted the need for enhanced security measures in academic institutions. The federal government grappled with significant third-party risk, with 28% of agencies exchanging data with over 5,000 parties. Meanwhile, the financial services sector consistently scored above all industries in risk assessments. These sector-specific challenges led to the development of targeted security frameworks and industry-specific best practices. SEE: Best CSPM Tools 2024: Top Cloud Security Solutions Compared Looking ahead: building cyber resilience Several key priorities

By Ivan Moreno ( December 20, 2024, 6:59 PM EST) — Insulet Corp. became the latest company to notch a colossal trade secrets award, and a new presidential administration has attorneys wondering what will become of the Federal Trade Commission’s pending proposal to ban employee noncompete agreements. Here’s a look at trade secrets cases that defined 2024 and what to expect from the FTC in the coming year…. Law360 is on it, so you are, too. A Law360 subscription puts you at the center of fast-moving legal issues, trends and developments so you can act with speed and confidence. Over 200 articles are published daily across more than 60 topics, industries, practice areas and jurisdictions. A Law360 subscription includes features such as Daily newsletters Expert analysis Mobile app Advanced search Judge information Real-time alerts 450K+ searchable archived articles And more! Experience Law360 today with a free 7-day trial. source

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More A startup founded by former Meta AI researchers has developed a lightweight AI model that can evaluate other AI systems as effectively as much larger models, while providing detailed explanations for its decisions. Patronus AI today released Glider, an open-source 3.8 billion-parameter language model that outperforms OpenAI’s GPT-4o-mini on several key benchmarks for judging AI outputs. The model is designed to serve as an automated evaluator that can assess AI systems’ responses across hundreds of different criteria while explaining its reasoning. “Everything we do at Patronus is focused on bringing powerful and reliable AI evaluation to developers and anyone using language models or developing new LM systems,” said Anand Kannappan, CEO and cofounder of Patronus AI, in an exclusive interview with VentureBeat. Small but mighty: How Glider matches GPT-4’s performance The development represents a significant breakthrough in AI evaluation technology. Most companies currently rely on large proprietary models like GPT-4 to evaluate their AI systems, a process that can be expensive and opaque. Glider is not only more cost-effective due to its smaller size, but also provides detailed explanations for its judgments through bullet-point reasoning and highlighted text spans showing exactly what influenced its decisions. “Currently we have many LLMs serving as judges, but we don’t know which one is best for our task,” explained Darshan Deshpande, research engineer at Patronus AI who led the project. “In this paper, we demonstrate several advances: We’ve trained a model that can run on-device, uses just 3.8 billion parameters, and provides high-quality reasoning chains.” Real-time evaluation: Speed meets accuracy The new model demonstrates that smaller language models can match or exceed the capabilities of much larger ones for specialized tasks. Glider achieves comparable performance to models 17 times its size while running with just one second of latency. This makes it practical for real-time applications where companies need to evaluate AI outputs as they’re being generated. A key innovation is Glider’s ability to evaluate multiple aspects of AI outputs simultaneously. The model can assess factors like accuracy, safety, coherence and tone all at once, rather than requiring separate evaluation passes. It also retains strong multilingual capabilities despite being trained primarily on English data. “When you’re dealing with real-time environments, you need latency to be as low as possible,” Kannappan explained. “This model typically responds in under a second, especially when used through our product.” Privacy first: On-device AI evaluation becomes reality For companies developing AI systems, Glider offers several practical advantages. Its small size means it can run directly on consumer hardware, addressing privacy concerns about sending data to external APIs. Its open-source nature allows organizations to deploy it on their own infrastructure while customizing it for their specific needs. The model was trained on 183 different evaluation metrics across 685 domains, from basic factors like accuracy and coherence to more nuanced aspects like creativity and ethical considerations. This broad training helps it generalize to many different types of evaluation tasks. “Customers need on-device models because they can’t send their private data to OpenAI or Anthropic,” Deshpande explained. “We also want to demonstrate that small language models can be effective evaluators.” The release comes at a time when companies are increasingly focused on ensuring responsible AI development through robust evaluation and oversight. Glider’s ability to provide detailed explanations for its judgments could help organizations better understand and improve their AI systems’ behaviors. The future of AI evaluation: Smaller, faster, smarter Patronus AI, founded by machine learning experts from Meta AI and Meta Reality Labs, has positioned itself as a leader in AI evaluation technology. The company offers a platform for automated testing and security of large language models, with Glider its latest advance in making sophisticated AI evaluation more accessible. The company plans to publish detailed technical research about Glider on arxiv.org today, demonstrating its performance across various benchmarks. Early testing shows it achieving state-of-the-art results on several standard metrics while providing more transparent explanations than existing solutions do. “We’re in the early innings,” said Kannappan. “Over time, we expect more developers and companies will push the boundaries in these areas.” The development of Glider suggests that the future of AI systems may not necessarily require ever-larger models, but rather more specialized and efficient ones optimized for specific tasks. Its success in matching larger models’ performance while providing better explainability could influence how companies approach AI evaluation and development going forward. source

TL;DR: The top-rated PDF Converter Pro can help you work more effectively with PDFs, and it’s on sale for 76% off through January 12. The business world runs on PDFs. Contracts are sent as PDFs, resumes are sent as PDFs, and internal documents are circulated as PDFs. After someone has made a document, they export it to PDF to make it as usable as possible for everybody else. However, if that recipient ever has to make changes, that’s when the problems start. PDFs are highly functional for protecting documents and making them more easily downloadable and printable, but they’re not particularly flexible unless you have a tool like PDF Converter Pro. This leading PDF software is now on sale for $76 off, and can revolutionize the way you work with PDFs. Features This all-in-one tool makes it significantly easier to work with PDFs. You can easily convert from PDF to Microsoft Word, Excel, PowerPoint, text, HTML, PNG and JPG for easy editing and viewing, and then convert back to PDF in just a couple of clicks. With integrated OCR technology, you can extract text from image-based PDF documents in the original format without any quality loss and ensure that all original layouts, images, texts, hyperlinks and more are preserved. In addition to converting, you can also merge and split PDFs to make them more practical to view, compress them to save space and password-protect any sensitive information. All of these features are why PDF Converter Pro has managed to earn a 4.4/5-star rating on Trustpilot. Streamline the way you and your business work with PDFs. Now through 11:59 p.m. PT on January 12, you can get a lifetime license to PDF Converter Pro for 76% off $99 at just $23.99. Prices and availability subject to change. source

By Alex Wittenberg ( December 20, 2024, 2:28 PM EST) — A Delaware bankruptcy judge on Friday signed off on the framework for Chilean mobile phone operator WOM SA’s $500 million restructuring plan, finding the debtor had exercised sound business judgment in selecting the deal to reduce some $650 million in debt…. Law360 is on it, so you are, too. A Law360 subscription puts you at the center of fast-moving legal issues, trends and developments so you can act with speed and confidence. Over 200 articles are published daily across more than 60 topics, industries, practice areas and jurisdictions. A Law360 subscription includes features such as Daily newsletters Expert analysis Mobile app Advanced search Judge information Real-time alerts 450K+ searchable archived articles And more! Experience Law360 today with a free 7-day trial. source

POS technology has impacted the retail industry far and wide ever since its debut in the 1970s. Before the modern POS system, transactions were manual, cash-based, and tangible. However, with the advent of technology, the POS has changed how retailers do business and how consumers do their shopping. Though the first POS was focused on processing cash payments in a clean, trackable way, it quickly evolved to facilitate credit and debit card transactions. And now? POS systems can do a whole lot more than just process payments. They’ve turned into mini business management command centers! So, what are the latest trends in POS systems? Below, I break down the POS retail trends of 2024 that are impacting the industry now and into the future. Cloud POS Business has transitioned to mostly cloud-based solutions — and that includes the POS. Cloud POS systems are increasingly becoming the norm, replacing the old-school traditional point-of-sale terminal that makes data difficult to access and manage. According to Straits Research, the global cloud POS market size was worth $3,987 million in 2022 and is forecasted to be more than $30,205 million by 2031 — that’s a compound annual growth rate (CAGR) of 24.7%. The cloud POS is a significant point of sale innovation​ because it has made data, agility, and mobility accessible to businesses of all sizes. Especially in a world where retail is no longer single-channel, cloud POS technology allows businesses to sync data and ensure operations run smoothly regardless of their business size or location(s). A retailer with 100 physical locations, its own website, and a thriving Amazon presence can benefit from the cloud POS just like the small mom-and-pop shop that’s managing a single retail location plus online channels. With cloud POS technology, you’re no longer tethered to a specific place or a hardwire connection. You can access your business technology from anywhere, at any time. AI and automation AI and automation are also changing virtually every aspect of the business landscape. I bet there are ways you already use these technologies that you don’t even realize! The trend is also impacting the POS — and in a good way. 100% of retail business owners surveyed by Square Future of Retail report say automation has improved their business in some way. That same survey found that tracking is the top area of automation for 43% of businesses, and the biggest benefit is an improved customer experience for 47% of businesses. Another 45% have seen improved employee retention and profitability thanks to automation. What role does AI play in modern POS systems? When it comes to POS technology, AI and automation happen in a ton of potential use cases: Automated reordering for inventory Fraud detection and prevention AI-powered pricing adjustments Loyalty and rewards programs Email marketing Shoppers from Square’s report say they’d support retailers automating: Checkout: 31% Product search: 22% AI-generated product descriptions: 22% Dynamic pricing: 21% Product recommendations: 18% Product reviews: 15% Seemingly, the possibilities are truly endless. I’m excited to see this trend play out and develop each year with innovative point-of-sale systems​. Payment options Today’s consumers like convenience and options. And this is true for payments made at the POS terminal. I remember when it was just a cash or credit world—now, customers can choose to pay for their purchases in many ways. According to Square, here’s how it breaks down in terms of the payment methods retailers accept: Cash: 58% Mobile wallet apps like Apple Pay or Google Pay: 57% QR code payments: 52% Traditional card payments: 47% Touchless card payments: 44% Buy-now, pay-later (BNPL) options like Clearpay: 43% POS technology is advancing in a way that accommodates these ever-changing payment preferences. In fact, modern POS trends show that the platforms have incorporated new features and functionality to process different payment methods and offer options like subscriptions, cryptocurrency, saved payments, bill splitting, and gift cards. Mobile POS Mobile POS, or mPOS, is another hot trend. The global mPOS market is estimated to be worth around $3.78 trillion, according to Statista. More than 2 billion mPOS users are expected by 2028. And nearly a quarter of shoppers want to use mobile checkout options. So why is this trending? Just like consumers are more mobile, so are associates. Now, associates can meet shoppers where they are on the sales floor and process the transaction on the spot. Customers don’t have to wait in line at the POS terminal — they can check out wherever they’re at. But this extends past transactions, too. An mPOS makes it easier for retail staff to perform various tasks, such as barcode scanning, inventory counts, etc. Customer self-serve Self-checkout is another trend in retail, and while it originally made its mark in grocery stores, other verticals are exploring the trend, too. This is largely consumer-driven — with Square Future of Retail reporting that two-thirds of shoppers prefer to do tasks independently, like checking out or seeing if a product is in stock. Consumers are increasingly feeling empowered to do things independently, influencing the future of POS. When asked what tasks they’d like to perform on their own with the help of technology as opposed to with a live staff member, here’s what the respondents said: Checking product inventory: 29% Ordering out-of-stock products: 26% Gathering information about a product: 26% Picking up an item ordered online: 24% Using mobile checkout or in-store terminals: 24% Arranging home delivery: 23% Why? They love the efficiency and convenience of being able to do things themselves. “Customers are looking for speed,” writes Bob Phibbs, The Retail Doctor. Self-checkout can provide that efficiency. These consumer preferences continue to impact point-of-sale market trends​. Biometric security Security should be a top concern for businesses — to protect both themselves and their customers. POS technology often comes equipped with security to safeguard all parties involved. PCI compliance, data encryption, and two-factor authentication are just a few examples. However, one security trend that will impact POS technology in 2025 and beyond is biometric security features.

❄ 🥂 🎇 🎀 🎄 🥳 Last year, I wrote a pastiche called the Twelve Days Of Loc[alization] Mess after a client asked about the root causes of localization problems. This week, I wondered if I should repost the blog, perhaps with a few updates — and was amazed to realize that, after one short year, I would change it almost totally. Last year’s loc-mess is still accurate and useful — sort of like a 201-level college course — but the conversations I’m having with clients and vendors have moved ahead substantially. Here is my updated version for the age of AI: On the twelfth day of loc-mess, my PM sent to me: 💻 Twelve all-English meetings 💁🏽‍♀️ Eleven stressed employees 🔨 Ten manual workflows ✅ Nine forms of QA 🤖 Eight unsafe LLMs 🌍 Seven language options 🏦 Six old-fashioned vendors 👨🏿‍💻 Five random TaaFs 📊 Four boring dashboards 🙋🏽‍♂️ Three confused execs 👩🏾‍💼 Two (too) few good leaders 🎯 And a non-audience-centric strategy. Your Loc-Mess Doesn’t Have To Be A Loc-Miss Any of these issues can reduce localization ROI. They have different root causes. And all of them are solvable. Let’s look at them one by one: All-English meetings. If you aren’t using AI to transcribe or dub your meetings, events, and videos into multiple languages, you are missing out on one of the fastest-growing and most transformative areas in localization. Transcribed subtitles are already an expectation for many audiences, and live dubbing, lip-synching, and translated, AI-generated summaries are becoming common. Stressed employees. For years, customer-facing employees and partners have struggled to communicate across language barriers. Sales and support staffing has been less flexible due to the need for language coverage. Now, businesses and governments are buying tools that enable multilingual live interactions, online chats, and emails — not just in one application but in all of them. When talking to clients, I’ve observed that when translation tools are available to everyone, adoption exceeds expectations by many multiples. Manual workflows. AI should automate every part of the localization workflow. If your teams are still submitting POs and doing manual file transfer; if translation occurs after development; or if your content teams are writing regional variations instead of using multilingual AI transcreation, you are missing out. Some of these tools are mature and others require more oversight, but the trend is overwhelming and accelerating. Linguistic quality assurance (LQA). Quality is not something you measure at the end — it’s something that you build into the entire process. In localization, that means cleaning up your content, translation memories, and glossaries; training neural machine translation (NMT) and large language models (LLMs); using linguistic quality evaluation (LQE) to select the best translation method; using auto-LQA for routine quality checks and humans for nuanced judgments; and continually improving MT/LLMs based on data. Unsafe LLMs. A BYOAI approach to translation is a terrible idea. Just like you don’t want your employees using public, unsecured, unbranded ChatGPT in your main language, you don’t want them using it to elevate risk in every language. Free translation tools are fine for foreign menus, not finance emails. Invest in trained, secure, paid LLMs for the whole company. Few supported languages. For decades, most B2B companies have translated into fewer than 10 languages, out of over 7,000 worldwide. Localization LLMs are changing this. Microsoft, Amazon, Google, DeepL, and others are expanding translation capabilities into hundreds of “language pairs” (e.g., English-French or Japanese-Thai), enabling companies to reach untapped markets. I predict that “EFIGS-JCK” support (English, French, Italian, German, Spanish, Japanese, Chinese, Korean) will become the mark of an old-fashioned business within a few years. Traditional vendors. The commodification of localization will elevate some vendors, eliminate others, and change almost all of them. This is the time to start treating your localization service and technology providers as strategic partners, not order-takers. Talk about their roadmaps. Ask how they will help you gain the benefits of AI while mitigating risk. Ask about new pricing and packaging models. Ask where human translators are still needed and where you can switch to NMT and LLMs. Look for vendors who are bold — and wise. Random TaaFs. Translation as a feature (TaaF) is showing up in applications everywhere: content management systems, chatbots, customer support systems, event software, meeting tools, etc. On the one hand, this is great. On the other hand, without central oversight, you’ll have inaccuracy, inconsistency, and inefficiency. If one group has already trained an LLM, then ask other vendors about leveraging it. Why pay twice for worse results? Boring dashboards. Localization is usually a revenue enabler, not a revenue generator, and it affects touchpoints throughout the customer and employee lifecycle. Measuring ROI and justifying investment is hard. The Forrester Balanced Scorecard for localization assesses audience experience and financial impact, agility, and operations while also showing how to talk to executives about localization impact in terms they’ll understand. Confused execs. When technology markets are undergoing rapid change, executives need clarity. Clients report that their decision-makers are asking: Do we still need human translators? Can we eliminate the localization budget now that TaaFs exist? Are NMT and LLMs the same thing, and are they both still relevant? Yes, no, no, and yes — but next year, the answers might change. Forrester’s Chart Your Course To Growth-Focused Localization helps business leaders create an aligned strategy over a 3–5-year horizon. Good leaders. More than ever before, you need expert localization leaders to guide your strategy and execution. It’s no longer adequate, if it ever was, to have disconnected localization efforts and technologies across functions and locales. Hire people who understand the details and implications of different solutions, giving them strategic oversight and accountability. Audience-centric strategy. None of the rest matters if you aren’t giving customers, partners, and employees what they need. Unless you have unlimited budget, localization prioritization is a tough nut to crack. Companies must look at touchpoints across the customer and employee lifecycle and assess preferences in each country and language. Forrester customers can take