If you're looking to subtract a percentage from a total in Excel based on the item name, you're not alone. Many users encounter challenges with functions like IF and SUMIF, especially when trying to automate calculations involving conditional adjustments. In your case, if the total of 2,200 reflects a 10% increase, you want the cell to display 2,000.

Introducing Residual Coupling (RC), a novel architecture designed for horizontally scaling language models without modifying weights. By connecting frozen models in parallel through learned linear bridges, RC enhances cross-domain generalization while preserving the integrity of base weights. This innovative approach stabilizes model outputs and mitigates overfitting, demonstrating significant improvements in perplexity and accuracy across various tasks.

In just 50 days, four significant supply-chain incidents involving OpenAI, Anthropic, and Meta have revealed a critical oversight in the security of release pipelines. These incidents, encompassing both adversary-driven attacks and self-inflicted failures, underscore a consistent gap in red-team evaluations, particularly around CI runners and packaging processes. With the emergence of the Mini Shai-Hulud worm and other vulnerabilities, the findings highlight the urgent need for AI vendors to rethink their security frameworks.

The Prove-It Economy is reshaping the marketing landscape, demanding transparency and accountability like never before. As brands strive to establish trust with consumers, many marketers find themselves unprepared for this shift. This article delves into the challenges and opportunities presented by the Prove-It Economy, highlighting the need for innovative strategies that prioritize authenticity.

Swiggy has enhanced its search autocomplete functionality by implementing a real-time machine-learning ranking system built on OpenSearch. This innovative architecture separates candidate generation from ranking, utilizing feature stores to incorporate real-time signals and applying learning-to-rank models to boost relevance. By replacing traditional heuristic ranking methods and adhering to strict latency constraints, Swiggy can continuously update its models based on user behavior.

In the fast-evolving field of AI engineering, navigating production trade-offs can be daunting, especially once your model is live. The article "Six Choices Every AI Engineer Has to Make (and Nobody Teaches)" delves into the critical decisions that often go unaddressed in formal training. By exploring these essential choices, you’ll gain insights that empower your engineering practices and enhance your project's success. For further reading on related technical challenges, check out "Building a Secure MCP Server on AWS for a Million-Company B2B Platform."

Elon Musk's lawsuit against Sam Altman and OpenAI has been dismissed, as a California jury unanimously ruled that Musk's claims of mistreatment by his cofounders were filed too late. This decision underscores the complexities surrounding one of the most talked-about figures in technology and his involvement with AI development. For those interested in the evolving landscape of artificial intelligence, explore our article "Top 10 AI Research Papers of 2025," which highlights significant advancements beyond traditional applications.

Anthropic's "Code with Claude 2026" in San Francisco showcased exciting advancements in AI technology, including the introduction of Managed Agents and Proactive Workflows within the Claude API platform. This event featured insightful discussions on developer experience, autonomy features, and the evolving impact of AI on product architecture, with contributions from industry leaders like GitHub and Vercel. For those interested in the latest trends in AI, our article, "Top 10 AI Research Papers of 2025," offers a deeper look into the significant shifts shaping the industry.

In this week's Java News Roundup for May 11, 2026, we delve into exciting developments including three OpenJDK JEPs aimed at JDK 27, the introduction of Azul Payara Community, and the WildFly wado CLI tool. Additionally, we’ll cover the latest point releases for LangChain4j and Google ADK, along with maintenance updates for Micronaut and OpenXava. For those interested in expanding their knowledge, check out our article on "Anthropic's Code With Claude," which highlights innovative approaches to managed agents and proactive workflows.

Grafana Labs, an open-source tool maker, has reported a security breach where hackers stole its codebase and threatened to release the source code unless a ransom is paid. This incident highlights the ongoing challenges faced by companies in safeguarding their intellectual property in an increasingly hostile digital landscape. For those interested in further exploring the implications of data breaches, check out our article, "NYC Health and Hospitals says hackers stole medical data and fingerprints during breach affecting at least 1.8 million people."

In the evolving landscape of data management, flexibility often triumphs over specialization. This article, "One Flexible Tool Beats a Hundred Dedicated Ones," explores why MCP servers struggle against CLIs once an agent gains terminal access. By embracing adaptable solutions, users can streamline workflows and enhance productivity. For those seeking deeper insights into the challenges faced by AI engineers in real-world applications, our piece "Six Choices Every AI Engineer Has to Make (and Nobody Teaches)" delves into critical production trade-offs.

Incorporating local language models into my daily workflow has revealed some intriguing insights. Contrary to common assumptions, I found that local models often outperform their cloud-based counterparts, proving to be not just viable options but superior choices. In this article, I’ll share five cool things I accomplished with local models, highlighting their unique advantages.

LangSmith Engine is transforming agent debugging by automating the identification and resolution of production failures, streamlining the entire process for AI engineers. By diagnosing root causes from the live codebase and drafting fixes in a single pass, it minimizes the time engineers spend addressing errors. While larger providers like OpenAI and Anthropic are integrating observability within their platforms, LangSmith offers a neutral layer that appeals to multi-model enterprises.

In a significant data breach, NYC Health and Hospitals has reported that hackers accessed the personal and medical information of at least 1.8 million individuals, including sensitive biometric data like fingerprints. This incident marks one of the largest breaches in 2026, raising serious concerns about data security in public healthcare systems. As the implications of such breaches unfold, organizations must prioritize innovative security measures to protect personal information.

Kin Health has successfully raised $9 million to develop an innovative AI notetaker designed specifically for patients. This app functions like a meeting notetaker, allowing users to record doctor visits and receive a concise AI-generated summary, complete with actionable next steps. Patients can easily share this information with family and friends, enhancing communication and understanding of their health journey. For more insights into the evolving landscape of healthcare technology, check out our article on the recent NYC Health and Hospitals data breach.

In the rapidly evolving landscape of enterprise AI, a staggering 95% of pilots fail to launch successfully. This post, "Why Your AI Demo Will Die in Production," delves into the critical factors behind this alarming statistic. It explores the complexities and challenges that arise once a demo transitions into a live environment, highlighting the pitfalls that can derail even the most promising initiatives.

Amazon's new Alexa+ feature empowers users to generate custom AI podcasts on demand, transforming Alexa into a personalized content platform. This innovation reflects Amazon's commitment to enhancing user experience through advanced technology, making content creation more accessible than ever. As the landscape of AI-driven tools evolves, exploring solutions like Alexa+ can redefine how we engage with audio content.

As the tech landscape evolves, the skills required for success in data roles are changing. This article explores the widening gap between what job candidates prepare for—primarily SQL and Python—and the dynamic needs of companies today. With many organizations seeking more versatile skill sets, it's essential to understand the additional competencies that can set candidates apart. For deeper insights into the challenges faced by AI engineers, check out "Six Choices Every AI Engineer Has to Make (and Nobody Teaches).

In 2025, AI research experienced transformative shifts, moving beyond traditional chatbots to focus on reasoning systems, autonomous agents, and multimodal approaches. Leading organizations like Google DeepMind, OpenAI, Anthropic, Meta, DeepSeek, and NVIDIA advanced the field with innovative papers on topics such as coding agents, reinforcement learning, and scalable safety systems. This exploration of new frontiers in AI sets the stage for exciting developments. For a deeper understanding of AI's practical applications, check out our article, "Why Your AI Demo Will Die in Production."

In this article, Shadi Elyafi explores the engineering challenges of building a secure Multi-Cloud Platform (MCP) server on AWS for a B2B intelligence platform that houses over a million company profiles. The goal is to enable users to query specific business criteria, such as "find SaaS companies in Germany with 50-200 employees," through a large language model (LLM) client. Key to this endeavor is ensuring a safe and effective workflow that protects production data while enhancing user accessibility and insights.

In the quest for seamless AI adoption, teams often unknowingly hinder their progress through miscommunication and resistance to change. Understanding the subtle ways these dynamics play out is essential for fostering a culture that embraces innovation. This article explores common pitfalls that teams encounter, offering practical insights to navigate these challenges effectively. For further insights into the broader implications of trust and technology, consider reading "Why trust is a big question at the Elon Musk-OpenAI trial," which highlights critical themes in AI development.

In your quest to enhance a searchable database, editing the results directly can be a transformative approach. By utilizing the filter function, you've populated your data effectively, but now you're looking to add a creative twist. Enabling edits in the results column not only streamlines your workflow but also allows for real-time updates to the database. This method can significantly improve data accuracy and accessibility.

Creating a Gantt chart in Excel can be a straightforward process, but the default template is limited to a 52-week timeline, which can be frustrating when managing long-term projects. If you've already invested time entering your project details and want to extend your chart to cover multiple years, there’s a way to modify the formulas for a more adaptable timeline. In this guide, we’ll explore how to adjust your Excel Gantt chart effectively while ensuring your project planning remains seamless.

Have you ever wondered why the formula `=RIGHT(PI())` returns 9 instead of the expected 4? This intriguing discrepancy stems from how different platforms interpret the `PI()` function. While Google Sheets displays the last digit of pi as 4, other systems may present it differently. Understanding these variations can enhance your spreadsheet skills and improve your data management. If you're curious about the nuances of data interpretation, check out our article "Would a new result in pre-print be considered by reviewers?" for more insights.

Join Stéphane Di Cesare and Cat Morris in "Product Thinking for Cloud Native Engineers" as they explore how engineers can transition from being viewed as a "cost center" to becoming valuable drivers of innovation through product discovery. They introduce the "Double Diamond" framework, emphasizing the importance of understanding user problems before developing solutions. Learn to select the right metrics, foster customer empathy through shadowing, and leverage business context to enhance the impact of your technical efforts.

I'm an undergraduate student from India thrilled to have been accepted to TAIGR, an ICML workshop where I'll be presenting a poster. However, I need financial aid for registration fees and accommodation, as my research is independent and I lack institutional backing. I would greatly appreciate any advice from those who have successfully applied for financial aid in the past. Your insights could be invaluable in navigating this process.

In this episode of our podcast, "Context is the Key to the Agentic Architecture Revolution," host Michael Stiefel engages with Baruch Sadogursky to explore the transformative role of software architecture in the era of agentic AI. They discuss how large language models (LLMs) can effectively reason through human ambiguity when supported by rigorous contextual artifacts. This approach allows software specifications to serve as a definitive source of truth, making the underlying code a flexible tool.

In this exploration, I investigate whether refusal layers in Mixture of Experts (MoE) models obscure safety failures influenced by dialect. Specifically, I compare responses to AAVE-coded prompts against those in Academic English during safety-sensitive scenarios. Key findings reveal significant behavioral divergences in model responses based on dialect, suggesting that routing differences occur before any explicit refusal. These insights raise critical questions about the adequacy of refusal mechanisms in addressing dialect-conditioned safety issues. For further insights, check out our article on the "Navigation API Reaches Baseline."

South Korea’s LetinAR is at the forefront of innovation, developing a thumbnail-sized lens that could serve as the optical backbone for the emerging era of AI glasses. This promising startup is redefining the possibilities of augmented reality, making advanced technology more accessible and user-friendly. As we explore how optical advancements can enhance our daily interactions with AI, consider reading "How teams accidentally sabotage AI adoption" for insights into overcoming barriers in technology integration.

Cloudflare and Stripe have unveiled an innovative protocol that allows AI agents to autonomously create cloud accounts, register domains, initiate subscriptions, and deploy applications into production. This groundbreaking development positions Stripe as a leader in managing identity and payment with a default cap of $100 per month, making it easier for organizations to harness AI capabilities. Notably, no other major cloud provider currently offers such agent-driven account provisioning.

The Navigation API has officially launched as a robust replacement for the History API, enhancing client-side navigation in single-page applications. Available in major browsers since January 2026, this new interface addresses previous limitations, offering a unified event model and improved history management. Key features include the navigate event, automatic URL updates, and integrated error handling, making it easier for developers to create seamless user experiences.

Unlock the full potential of OpenAI's Codex with our guide on maximizing this innovative coding agent. Learn how to streamline your coding processes, enhance productivity, and tackle complex tasks with ease. By harnessing Codex's capabilities, you can transform your programming experience and elevate your projects. For those interested in related topics, check out our article on creating an Excel Gantt chart, which offers insights into project planning and task management. Dive in and discover how Codex can empower your coding journey.

Introducing the model-agnostic sensitivity approximator, a tool designed to enhance explainable AI by evaluating how sensitive model predictions are to individual features. While existing tools like SHAP and LIME focus on feature attribution, this package takes a step further, offering insights into effective risk management for black box models, including random forests and XGBoost. By employing a perturbation-based approach, it provides stable sensitivity estimates, particularly beneficial when gradients are not analytically available.

In the context of peer review, the discovery of an updated preprint on arXiv can significantly influence your evaluation of a paper. If the authors have addressed a critical oversight—like an unexamined theoretical result or experiment—it raises the question: should you still critique the original submission for this omission? Your decision might hinge on whether you believe the authors will incorporate these updates in their final manuscript. For further insights into data management challenges, consider exploring our article on "Filtering with conditions and pivot table."

Welcome to this week's Entering & Transitioning thread! Here, you can ask questions about starting or advancing your journey in the data science field. Whether you're seeking learning resources, exploring traditional or alternative education options, or navigating job search challenges, this thread is your space for guidance. While you await community responses, feel free to check out the FAQ and Resources pages on our wiki for immediate support. For deeper insights, explore our article on "Filtering with conditions and pivot table" to enhance your skills.

When reflecting on your daily work activities, consider how much enjoyment and intellectual stimulation they provide. Are the thought processes and problem-solving aspects of your role satisfying, or do they feel lackluster? For many, the distinction between a fulfilling career and "just another white-collar job" can hinge on these experiences. If you’re new to the field and seeking insights, exploring discussions like "Weekly Entering & Transitioning - Thread 18 May, 2026 - 25 May, 2026" can provide valuable perspectives on navigating your journey.

Filtering loans based on collateral types can streamline your data analysis. To identify loans with only one type of collateral, such as vehicle collateral, you can utilize a pivot table alongside a COUNTIFS formula. For loans secured by multiple collateral types, the challenge lies in ensuring that loans like DEF/456 aren’t miscounted when filtering for property collateral. A specific formula can help you refine this filtering process.

If you're looking to remove highlighted boxes from your spreadsheet, you're not alone. Many users encounter this challenge and may initially try using the “no fill” option to eliminate the highlights. However, there are additional steps you can take to achieve a cleaner look. Explore our guide for practical solutions that simplify the process. For further insights, check out our article on handling training data when public datasets don’t match your needs, which can enhance your overall spreadsheet experience.

Explore the latest advancements in LLM architectures, including key developments in KV sharing, mHC, and compressed attention. These innovations are shaping the future of machine learning, making complex processes more efficient and accessible. Understanding these concepts can empower you to leverage AI technology effectively in your projects. For further insights, you may find our article, "How are you handling training data when public datasets don't match your use case?" particularly relevant as it addresses challenges in adapting data to fit specific needs.

The ongoing trial involving Elon Musk and OpenAI has brought the question of trust to the forefront, particularly regarding OpenAI CEO Sam Altman's credibility. As the trial progresses, the scrutiny of Altman’s actions and decisions highlights the broader implications of trust in leadership within the tech industry. This theme resonates with current concerns about privacy and data management, as seen in our article on Apple's anticipated Siri revamp, which emphasizes the importance of user trust in technology.

Navigating the challenge of training data when public datasets fall short can be daunting. Many teams face the dilemma of either accepting degraded performance with existing data, investing weeks in scraping and cleaning, or employing augmentation techniques that offer limited improvement. However, there’s an innovative approach that involves sourcing permissively licensed real-world data, curating it to fit specific schemas, and applying synthetic expansion to enhance volume and coverage. If you're encountering data bottlenecks in your projects, I'd love to hear about your experiences.

If you’re encountering #NAME? errors in Excel 365 when opening a workbook, you’re not alone. Many users face this frustrating issue, especially when formulas only work after being manually confirmed in the formula bar. Unfortunately, refreshing with Ctrl+Alt+Shift+F9 may not resolve the problem. Fortunately, there are faster methods to force calculations across thousands of rows. For a deeper understanding of optimizing your Excel experience, consider exploring our article on “Formatting templates for finance” to enhance your workflow efficiency.

OpenAI has open-sourced Symphony, an innovative agent orchestrator designed to streamline autonomous coding. By leveraging project-management tools like issue trackers, Symphony efficiently coordinates multiple coding agents, each dedicated to completing specific tasks without direct developer oversight. This approach enhances productivity, allowing human reviewers to focus on the quality of the output instead of managing interactive sessions. For those looking to optimize their Excel experience, explore our article "Formulas are returning #NAME? errors on opening workbook in Excel 365" for practical insights on troubleshooting common issues.

Setting up a formatting template in Excel can streamline your workflow and enhance clarity. For your needs, you can assign specific colors to different types of inputs: hardcodes in blue, formulas in black, and user inputs in red. This visual differentiation can help you quickly identify data types and maintain accuracy in your financial analyses. If you’re looking for more insights on data management, consider checking out "Formula for data sets and differences," which delves into effective formula creation and data comparison techniques.

In the rapidly evolving landscape of AI investments, understanding the key factors that distinguish successful ventures from failures is essential. This article explores five critical levers that can enhance your investment strategy, ensuring you make informed decisions in this dynamic market. By focusing on these elements, you can better navigate the complexities of AI technology and its applications. For further insights, consider reading our related piece, "How to Analyze Company Earnings with AI in 2026," to deepen your understanding of AI's impact on financial analysis.

Apple's upcoming revamp of Siri is set to emphasize privacy, potentially introducing features like auto-deleting chats to enhance user security. This shift reflects a growing demand for more responsible data management in AI technology. As Apple unveils this new version, users can expect a more streamlined and secure experience. For those interested in exploring the evolving landscape of AI interactions, check out our article "Agents vs Chatbots: Codex Changes Everything" for insights on how AI is reshaping digital communication.

Welcome back to TechCrunch Mobility, your go-to source for insights on the evolving landscape of transportation. As the automotive industry gears up for an AI skills arms race, we’re witnessing a pivotal shift in how technology shapes our vehicles and driving experiences. This transformative journey invites exploration into innovative solutions that redefine mobility. For those curious about the broader implications of AI, check out our article, “If you’re giving a commencement speech in 2026, maybe don’t mention AI,” for a thought-provoking perspective on the future.

Eclipse's recent $2.5 billion investment in Cerebras marks a pivotal moment in its journey to realize its physical-world thesis. Once a lonely endeavor for Lior Susan a decade ago, this strategic move positions his firm at the forefront of the tech industry's evolution. With a renewed focus on real-world applications, Eclipse is not just navigating the landscape but shaping it. For further insights into the intersection of technology and innovation, be sure to check out our article on the AI skills arms race in automotive.

Creating a formula to compare two data sets and highlight differences can streamline your data analysis and improve clarity. By combining these differences in parentheses, you can easily visualize changes and insights. For a deeper dive into similar challenges, check out our article on calculating averages with multiple criteria, titled "Got a table with several names that repeat and values to them, I need to calculate the average of the values of 3 names only." Let’s explore how to enhance your spreadsheet capabilities together.

If you're looking to bridge the gap between theoretical knowledge and practical application of machine learning algorithms, you're not alone. Many learners seek guidance on which algorithms suit specific datasets and how to implement them step by step. To enhance your understanding, consider exploring curated resources that break down these concepts into actionable insights. For a deeper dive into recent advancements, check out our article on "Recent Developments in LLM Architectures," which discusses innovative techniques that can enrich your learning journey.