Technology

What if the most important AI device at the FIFA World Cup isn’t a camera—but the ball itself? ⚽🤖 Modern FIFA match balls contain embedded sensors that capture hundreds of data points every second. Combined with AI-powered computer vision, they help determine the exact moment the ball is played, improving offside decisions, enhancing VAR, and generating rich real-time match analytics. But this is bigger than football. It’s a powerful example of sensor fusion—where edge devices, AI models, and high-performance computing work together to deliver insights in milliseconds. The same principles are transforming industries far beyond sport: 🏭 Smart manufacturing with connected sensors. 🚗 Autonomous vehicles combining cameras, radar, and LiDAR. 🏥 Healthcare devices delivering real-time diagnostics. 🏙️ Smart cities optimizing traffic, energy, and public safety. As AI moves from the cloud to the edge, every connected device becomes a source of intelligence. Turning that data into instant, reliable decisions requires powerful compute infrastructure. This is where AMD is helping drive the next era of AI—from Ryzen AI PCs at the edge, to EPYC processors powering modern data centers, and Instinct accelerators enabling large-scale AI inference and training. The future of AI isn’t just about bigger models. It’s about connecting billions of intelligent devices with the compute needed to make every decision count—in real time. From the football pitch to the factory floor, AI is changing how the world works. #AI #AMD #EdgeAI via @untoldoddities #SportsTech #ComputerVision #SensorFusion #HighPerformanceComputing #DataCenter #DigitalTransformation #Innovation #FIFA

This is the most underrated way to use Claude: (and it has nothing to do with writing or coding) It's competitive intelligence. Using data that's free, public, and updated every single week. Here's my extract step by step guide: Step 1. Go to claude .ai. Step 2. Select the new Claude "Opus 4.6." Step 3. Turn on "Extended Thinking." Step 4. Pick a competitor. Go to their careers page. Step 5. Copy every open job listing into one doc. (Title. Team name. Location. Full description) Step 6. Save it as one .txt or .docx file. Step 7. Search the company at EDGAR (sec .gov) Step 8. Download its recent 10-K or 10-Q filing. (Official strategy, risks, and financials - all public.) Step 9. Upload both files to Claude Opus 4.6. Step 10. Paste this exact prompt: "You are a competitive intelligence analyst at a rival company. I've uploaded [Company]'s complete current job listings and their most recent SEC filing. Perform a strategic intelligence analysis: → Cluster these roles by what they suggest is being built. Don't use the team names they've listed. Infer the actual product initiatives from the skills, tools, and responsibilities described. → Identify capabilities or teams that appear entirely new — not mentioned anywhere in the SEC filing. These are unreleased bets. → Find roles where seniority is disproportionately high for a new team. This signals executive-level priority. → Cross-reference the SEC filing's Risk Factors and Strategy sections with hiring patterns. Where are they investing against a stated risk? Where did they flag a risk but have zero hiring to address it? → Predict 3 product launches or strategic moves this company will make in the next 6-12 months. State your confidence level and cite specific job titles and filing sections as evidence. Format this as a 1-page competitive intelligence briefing for a CMO." What you'll find: → Products that don't exist yet but will in 6 months. → Priorities that contradict what the CEO said. → Risks they told the SEC but aren't addressing. This is what consulting firms charge $200K for. It took me 10 minutes. I used the new Claude 'Opus 4.6' for a reason: ✦ It read 60 job listing & a 200-page filing together.  ✦ And connects dots across both. ✦ It is superior in thinking and context retrieval. That's why I didn't use ChatGPT for this.

Every cloud provider faces the same AI infrastructure challenge: chips need to be positioned close together to exchange data quickly, but they generate intense heat, creating unprecedented cooling demands. We needed a strategic solution that allowed us to use our existing air-cooled data centers to do liquid cooling without waiting for new construction. And it needed to be rapidly deployed so we could bring customers these powerful AI capabilities while we transition towards facility-level liquid cooling. Think of a home where only one sunny room needs AC, while the rest stays naturally cool – that’s what we wanted to achieve, allowing us to efficiently land both liquid and air-cooled racks in the same facilities with complete flexibility. The available options weren't great. Either we could wait to build specialized liquid-cooled facilities or adopt off-the-shelf solutions that didn't scale or meet our unique needs. Neither worked for our customers, so we did what we often do at Amazon… we invented our own solution. Our teams designed and delivered our In-Row Heat Exchanger (IRHX), which uses a direct-to-chip approach with a "cold plate" on the chips. The liquid runs through this sealed plate in a closed loop, continuously removing heat without increasing water use. This enables us to support traditional workloads and demanding AI applications in the same facilities. By 2026, our liquid-cooled capacity will grow to over 20% of our ML capacity, which is at multi-gigawatt scale today. While liquid cooling technology itself isn't unique, our approach was. Creating something this effective that could be deployed across our 120 Availability Zones in 38 Regions was significant. Because this solution didn't exist in the market, we developed a system that enables greater liquid cooling capacity with a smaller physical footprint, while maintaining flexibility and efficiency. Our IRHX can support a wide range of racks requiring liquid cooling, uses 9% less water than fully-air cooled sites, and offers a 20% improvement in power efficiency compared to off-the-shelf solutions. And because we invented it in-house, we can deploy it within months in any of our data centers, creating a flexible foundation to serve our customers for decades to come. Reimagining and innovating at scale has been something Amazon has done for a long time and one of the reasons we’ve been the leader in technology infrastructure and data center invention, sustainability, and resilience. We're not done… there's still so much more to invent for customers.

Two strikingly similar headlines surfaced this past week that should make every leader pause: • “Companies Are Pouring Billions Into A.I. It Has Yet to Pay Off.” — New York Times • “Companies Are Pouring Billions Into AI. Here’s Why They’re Not Seeing Returns” — Forbes The NYT points to the human side: employees resist tools they don’t trust. Forbes focuses on the technical side: most AI still can’t understand the context of work. Both are true, and they’re related. When AI lacks context, employees lose trust. It can’t tell the latest doc from last year’s draft. It summarizes a customer conversation but drops the follow-ups buried in the thread. It pulls a response from Slack while ignoring the context in Google Drive. Employees realize it creates more work than it saves, and stop using it. Pilots stall, deployments fade, and projects slide into the “trough of disillusionment" as the NYT describes. Unfortunately, that's the reality for many organizations. At Glean, we work hard to make sure AI understands the enterprise context the way a human does. If a subject matter expert says something, I trust it more. If something’s old, I double-check it. That’s how people think, and it’s how AI should work too. Yet every enterprise has its own documentation culture and quirks, so sometimes we struggle at first. But we persist and co-develop with customers until the system reaches the quality they need. Then we take those learnings to make it work automatically for the next customer. We’ve seen this approach deliver measurable impact for customers: • Booking.com: Glean Agents give teams faster access to customer insights, cutting video production time by 75% and doubling monthly output. • Confluent: Glean’s AI-powered search saves 15,000+ hours/month, boosts support satisfaction by 13%, and cuts ticket investigation time by 10 minutes. • Fortune 100 telecom company: Glean surfaces instant knowledge during support calls, reducing call resolution time by 17 seconds across 800+ agents. • Leading global consultancy: Glean Agents automate RFP workflows, cutting consulting project proposals from 4 weeks to a few hours (97% faster). • Wealthsimple: Glean gives employees instant access to policies and knowledge, driving $1M+ in annual productivity gains. When AI understands the real context of work—across people, tools, and workflows— employees trust it and use it. Instead of falling into the trough of disillusionment, companies climb a slope toward productivity gains and real ROI.

The 𝐂𝐡𝐢𝐞𝐟 𝐃𝐢𝐠𝐢𝐭𝐚𝐥 𝐎𝐟𝐟𝐢𝐜𝐞𝐫 (𝐂𝐃𝐎) role emerged in the early 2000s as companies recognized the need for a dedicated executive to lead their digital transformation efforts. The role was created to help organizations navigate the rapidly changing digital landscape and effectively integrate technology into their overall business strategy. According to PwC, 𝟐𝟏% of public firms now have a CDO, up from 𝟔% in 2014. Yet, according to NewVantage Partners | A Wavestone Company, 𝟔𝟓% of data-intensive firms have a CDO in place. This data shows the importance and urgency that companies are placing on data and managing a digital strategy. Yet, at the same time, the process has been slow: • Only 𝐡𝐚𝐥𝐟 of CDOs are able to drive innovation using data. • Just 𝟒𝟎% of CDOs manage data as a business asset. • Roughly 𝟐𝟔% of CDOs have succeeded in creating a data-driven organization. • About 𝟐𝟓% of CDOs have no single point of accountability for data within their organization. • Only 𝟒𝟎% of respondents said the CDO role is successful and established within their organization. 𝐖𝐡𝐲 𝐝𝐨 𝐬𝐨 𝐦𝐚𝐧𝐲 𝐂𝐃𝐎𝐬 𝐟𝐚𝐢𝐥? 𝟏. 𝐋𝐚𝐜𝐤 𝐨𝐟 𝐫𝐨𝐥𝐞 𝐝𝐞𝐟𝐢𝐧𝐢𝐭𝐢𝐨𝐧 – Anytime a new type of role is created in the industry, everyone is bound to have a different interpterion of what it means. As this role is often designed to be disruptive, it is important to establish clear responsibilities, priorities, and boundaries. 𝟐. 𝐋𝐢𝐦𝐢𝐭𝐞𝐝 𝐚𝐮𝐭𝐡𝐨𝐫𝐢𝐭𝐲 𝐚𝐧𝐝 𝐫𝐞𝐬𝐨𝐮𝐫𝐜𝐞𝐬 – Without adequate resources and authority, CDOs may be unable to effectively execute their digital strategy and drive digital transformation within the organization. 𝟑. 𝐓𝐨𝐨 𝐡𝐢𝐠𝐡 𝐨𝐟 𝐞𝐱𝐩𝐞𝐜𝐭𝐚𝐭𝐢𝐨𝐧𝐬 – Digital transformation initiatives are almost always massive changes for a company. CDOs often face resistance from employees who are resistant to change and unwilling to adapt to new technologies and ways of working. 𝟒. 𝐃𝐢𝐟𝐟𝐢𝐜𝐮𝐥𝐭𝐲 𝐢𝐧 𝐦𝐚𝐧𝐚𝐠𝐢𝐧𝐠 𝐨𝐫𝐠𝐚𝐧𝐢𝐳𝐚𝐭𝐢𝐨𝐧𝐚𝐥 𝐬𝐢𝐥𝐨𝐬 – CDOs can struggle to bridge the gap between different departments and break down silos that prevent collaboration and information sharing. 𝟓. 𝐋𝐢𝐦𝐢𝐭𝐞𝐝 𝐄𝐱𝐞𝐜𝐮𝐭𝐢𝐯𝐞 𝐚𝐜𝐜𝐞𝐩𝐭𝐚𝐧𝐜𝐞 – Whenever you create a new role for the first time, you’re ultimately taking away an area of responsibility of someone else, usually the Chief Information Officer (CIO) in this case. Is this new position welcomed or feared by other executives? 𝐒𝐭𝐚𝐭𝐢𝐬𝐭𝐢𝐜𝐬 𝐒𝐨𝐮𝐫𝐜𝐞𝐬: https://lnkd.in/eu3aiHnF https://lnkd.in/eNeERKmb ******************************************** • Follow #JeffWinterInsights to stay current on Industry 4.0 and other cool tech trends • Ring the 🔔 for notifications!

🚛 WHEN TRANSPORT LEARNS TO THINK GREEN I came across a concept today that stopped me — an autonomous hydrogen truck-trailer drone designed for long-distance freight. At first, it looked like another futuristic vehicle. But then it hit me: this isn’t just transport evolving — it’s intent evolving. For decades, we’ve designed logistics around speed and scale. Now we’re finally designing around sustainability. This new concept merges autonomy, aerodynamics, and hydrogen power to do something radical: → Eliminate carbon emissions in heavy freight. → Cut operational energy costs through intelligent routing. → Reduce highway congestion with coordinated drone convoys. It’s not just engineering — it’s a shift in philosophy. A move from moving faster to moving responsibly. We often talk about “green tech” as a feature — but the real shift happens when sustainability becomes the invisible infrastructure behind innovation. It’s not an addition to progress. It is progress. What’s needed now isn’t more invention — it’s integration. We need to: ✅ Build networks where clean energy and automation reinforce each other. ✅ Redefine “efficiency” to include environmental balance. ✅ Shift from carbon offsetting to carbon prevention at design level. Because the next breakthrough won’t come from faster engines — but from systems that make waste impossible by design. That’s when technology stops being an experiment in innovation… and becomes an expression of intelligence. So here’s the question I keep returning to — 👉 Will the next era of transport be powered by fuel — or by foresight? #Innovation #Sustainability #Hydrogen #AutonomousVehicles #GreenTech #Logistics #FutureThinking

I shared a new tutorial + experiments on finetuning LLMs for classification efficiently. In this video, I explain how to convert a decoder-style LLM into a classifier. Many business problems are text classification problems, and if classification is all we need for a given task, using "smaller" and cheaper LLMs makes a lot of sense! (But, of course, also always run a simple logistic regression or naive Bayes baseline to determine if you even need a small LLM.) 🧪 In addition, I also ran a series of 19 experiments to answer some "what if" questions around finetuning pretrained LLMs for classification. Here, I kept things simple and small (e.g., GPT-2 on a toy binary classification task): Here's a snapshot summary of some of the interesting ones: 1) As would be expected, training on the last token yields much better performance than the first 2) Training the last transformer block is way better than just the last layer 3) LoRA performs on par or better than full finetuning—while being faster and more memory-efficient 4) Padding to full context length hurts performance 5) No padding or smart position selection leads to consistently higher accuracy 6) Surprisingly, training from random weights isn't much worse than using pretrained 7) Averaging embeddings over all tokens can improve performance slightly with little cost The full video is available here: https://lnkd.in/gcfqR2mH PS: If you are wondering why GPT instead of BERT? Well, you can of course also use BERT. Based on experiments on the 50k Movie Review dataset It's interesting though that this 3x smaller LLM performs on par (actually slightly better) than BERT. (ModernBERT then again is 2% better.)

Basics of Cybersecurity: What Every Tech Professional Must Know Today In our world, cybersecurity knowledge isn't optional anymore. Let me share some actual numbers and practical insights that matter to every Tech professional: The Big Three Threats You Need to Know: 1. Phishing attacks cause 90% of all data breaches. These aren't just spam emails - they're sophisticated scams that can fool even experienced users. The fix? Strong email filters and two-factor authentication are your best defense. 2. Ransomware isn't just about paying ransom - companies lose millions in downtime alone. Regular backups and solid recovery plans are essential, not optional. 3. DDoS attacks can shut down your entire business in minutes. Cloud-based protection and load balancing aren't fancy extras - they're basic necessities. What has really worked in 2024: - End-to-end encryption for all sensitive data - Regular security training for all staff (not just IT) - Automated threat detection tools - Continuous system monitoring The Truth: Most successful attacks exploit basic security gaps. Good security isn't about complex solutions - it's about getting the fundamentals right every single day.

Alignment without context integrity is not safety. An AI agent can faithfully follow its instructions and still be dangerous if its understanding of reality is wrong. Anthropic recently disclosed that Claude models gained unauthorized access to the real systems of three organizations during cybersecurity evaluations. The agents had been told they were operating in a simulation with no internet access. But a configuration mistake gave them access to the live internet. They treated real production systems as part of the exercise and kept pursuing the goal they had been given. OpenAI separately disclosed that models found a previously unknown vulnerability, escaped an isolated evaluation environment and compromised Hugging Face. These were not simply failures of intelligence. The deeper problem was that the agents were acting inside a false understanding of reality. We have spent years asking whether an AI system will follow our instructions. We now also need to ask whether it correctly understands the environment in which those instructions are being executed. This creates a new security requirement. Context integrity. Before an agent acts, the system must continuously verify where it is, which resources are in scope, whose authority it carries, what it is allowed to do, and when that authority expires. Just in time permission for every action. At just the right time. For just enough time. Assessed in real time. Those facts cannot live only inside a prompt. They must be verified and enforced by the infrastructure around the model. A prompt is not a security boundary. Zero trust taught us to never trust identity and always verify access. And provide least privileged access. Agentic AI adds another dimension. Never blindly trust context. Continuously verify reality. The next security perimeter is not just the agent’s identity. It is the agent’s understanding of reality. The most dangerous agent may not be misaligned. It may simply be mistaken. And in an agentic world, a false belief can become a real breach.

Billions of people unlock their phone with their face every day. Apple Face ID. Google face unlock. Neither would pass the identity standard the EU is about to enforce 😳 And here's the uncomfortable part: the verification most banks use today wouldn't either. Document-based identity verification was never designed for the digital world. Passports have UV and infrared security features that can only be checked physically. Online, you're verifying a photograph of a secure document. A photograph. That was fine when faking an identity required skill, time, and resources. In 2026, AI does it in seconds. → Synthetic identities pass document capture → AI-generated faces clear liveness checks → Deepfakes defeat the "blink twice, turn your head" routine → Full identity packages are assembled faster than compliance teams can update their rules Verifying a document no longer means you've verified a person. And that's a fraud problem, not just a compliance one. Governments see it. That's why they're not patching the old model - they're replacing it with digital-native identity. Government-backed eIDs that are cryptographically assured, not photographed. You're not trusting a scan. You're trusting a digitally signed identity that AI can't fake. And regulation is now forcing the pace. eIDAS 2.0 is law. The AML Regulation update hits in July 2027. Banks, insurers, and telecoms are being pushed toward higher-assurance identity - fast. But 150+ eID schemes exist worldwide, each with different standards. Integrating them individually doesn't scale. That's what Hopae is building - a global eID network connecting 100+ government-backed identity schemes through a single integration. One API. Access to the highest-assurance digital identities globally. This is how fraud actually goes down. Not by adding another layer to a broken process, but by replacing the process entirely. If you want to see how this is being built in practice, take a look here: https://lnkd.in/dW9pa6wV