IBM's Z Atlas AI Maps Entire z/OS Mainframe Stack

Key Takeaways

• <ul><li>IBM's Z Atlas AI tool maps the full z/OS stack to simplify mainframe modernization</li><li>HCLTech launched AI agents for mainframe operations with human oversight intact</li><li>Johns Hopkins' AstroPath cancer mapping tech was selected for Anthropic's AI for Science program</li><li>Ralph Debbas, W Motors founder, appointed Group President of Robo.ai</li><li>MIT CSAIL developed Weight-Space Programming allowing AI to read and rewrite neural network weights</li><li>Intel proposes Edge, Agentic, and Physical AI layers for cognitive cities</li><li>Berlin hosted IFA 2026 with 240,000 visitors and a government AI summit</li><li>Norway survey finds petroleum industry AI use needs stronger oversight</li><li>AWS developer connected AI coding agent Kiro to a vintage Teletype machine</li><li>84% of cities worry about unexpected AI costs according to ThoughtLab study</li></ul>

IBM launches AI tool Z Atlas to simplify mainframe modernization

IBM is developing Z Atlas, an AI-powered tool designed to help companies understand and modernize their mainframe systems. The tool maps the entire z/OS stack across infrastructure, middleware, and applications. It aims to reduce manual work for tasks like patch management and knowledge sharing. Z Atlas is currently in a proof of concept phase and creates a shared intelligence layer for teams to manage IBM Z environments using natural language.

HCLTech launches AI agents for mainframe operations

HCLTech has launched an AI-driven solution to automate and improve mainframe operations. The offering uses specialized mainframe skill agents and conversational AI assistants to support system administration and monitoring. It is backed by HCLTech's Mainframe Center of Excellence. The solution aims to help companies manage critical systems proactively while keeping human oversight in place.

AI could make emotional intelligence more important at work

AI systems can recognize patterns in language and sound empathetic, but they do not truly experience emotions. Experts say this makes human emotional intelligence more important as companies use AI in more work tasks. Leaders must decide where to use AI and how to trust its outputs. Skills like self-awareness, empathy, and critical thinking help people judge AI results objectively and manage real workplace situations.

Johns Hopkins cancer mapping tech chosen for Anthropic AI project

Johns Hopkins University researchers had their AstroPath cancer mapping technology selected for the Anthropic AI for Science program. AstroPath uses technology from astronomy to create detailed maps of tumors. The program provides free access to Anthropic's Claude AI models to help analyze complex data. The team uses AI to speed up tasks that once took researchers months of manual work.

AI in petroleum industry needs better oversight, says survey

A survey by the Norwegian Ocean Industry Authority found that AI use in the petroleum industry is growing but needs stronger management. Companies have different levels of maturity in handling AI risks and oversight. Some companies do not provide enough information about how they use AI. The survey recommends clear AI policies, employee training, and regular monitoring and reporting on AI use.

Berlin hosts digital summit focused on AI risks and uses

Germany is hosting a government-run conference on artificial intelligence in Berlin. The event will explore both the benefits and dangers of AI. German and UK defense ministers will also meet to discuss expanding AI use in military cooperation. The summit highlights growing international interest in shaping AI policy.

Developer connects AI coding agent to vintage teletype machine

AWS developer advocate Darko Mesaroš connected an AI coding assistant called Kiro to a retro Teletype terminal. He used a WiModem232 for the serial connection and a custom Rust app to format output for the old device. The project explores the Agent Client Protocol that powers Kiro. Although it serves no real purpose, the experiment teaches serial connectivity and ACP protocol skills. Mesaroš said the setup could work with any coding agent that supports ACP.

Robo.ai names W Motors founder Ralph Debbas as Group President

Robo.ai Inc. appointed Ralph R. Debbas as Group President on November 2, 2026. Debbas founded W Motors in 2012 and led development of the Lykan Hypersport supercar. He will focus on industrial strategy and international partnerships for Robo.ai's AI and advanced manufacturing businesses. The company is reaching milestones including ALIF Holding's industrial park groundbreaking and Neurovia AI's NeuroStream platform launch.

AI and robotics dominate the floor at IFA 2026 in Berlin

IFA 2026 in Berlin showcased AI-powered appliances, robotic companions, and mobility exoskeletons. Over 240,000 visitors from 155 countries attended the consumer electronics show. Standout products included the Hypershell Halo exoskeleton and Hisense's generative AI companion suite. The global consumer electronics sector grew 16.6 percent since 2021 despite a five percent slowdown in China.

Intel outlines AI strategy to make cities more responsive and adaptive

Intel Vice President Renu Navale described how AI can transform cities from smart to cognitive. The approach follows a loop of Sense, Understand, Decide, Act, and Learn. Intel proposes Edge AI, Agentic AI, and Physical AI as three key layers. A ThoughtLab study shows 84 percent of cities worry about unexpected costs and 82 percent lack AI skills. Sarasota, Florida already uses computer vision with traffic cameras to manage congestion.

MIT Researchers Create Weight-Space Programming for AI

MIT CSAIL researchers developed a new method called Weight-Space Programming that lets AI models read and rewrite the numerical weights of other neural networks. This technique allows an AI to understand what a network does just by looking at its raw numbers without seeing the training data. In tests, the AI correctly identified which games four networks played and what language three language models used based only on their internal weights. The system also wrote working code for game-playing networks from scratch, beating traditional training methods that took millions of steps. This breakthrough shows that AI can fix or improve existing models by directly editing their numbers instead of retraining them.

Sources

• IBM aims to simplify mainframe modernization with AI-powered insight
• HCLTech launches AI agents to automate enterprise mainframe operations
• AI Could Make Emotional Intelligence More Important, Not Less
• Johns Hopkins cancer cell mapping technology chosen for Anthropic AI project
• Artificial intelligence (AI) in the petroleum industry is in need of better oversight and management
• Germany news: Berlin hosts digital summit with AI focus
• Vintage fun: Prodding AI agents using retro hardware just because you can
• Robo.ai Appoints W Motors Founder Ralph R. Debbas as Group President
• AI and autonomous hardware takes centre stage at IFA 2026
• Cognitive Cities: How AI Can Help Cities Move From Seeing Conditions...
• AI Models Can Understand And Improve Model Weights Through “Weight-Space Programming,” Says MIT Research