AME Agent Swarms Quietly Rewrite the software development life cycle Workflow at AMD

Amazing stuff! Impressive!

"The impact of AI on software development has been both profound and ever-evolving. Last year, I wrote about AMD’s plans to use AI not just for generating new lines of code, but also for other steps in the software development lifecycle (SDLC), such as triaging problems, debugging code, and testing the software. At the time, we were hoping for a 25 percent productivity boost from AI use over the course of two or three years.

But with each new release, the capabilities of large language models (LLMs) improve dramatically—accelerating software development, increasing the quality of AI-generated code, and fundamentally reshaping how software is engineered.

Now, just one year later, we have surpassed our productivity target, achieving a 30 percent overall productivity boost through AI. On top of that, we are rethinking not only how we use AI within the SDLC, but the structure of the SDLC itself. ...

AMD began developing AI systems for code generation, testing automation, bug analysis, and code review in 2024. At the time, our objective was to achieve 25 percent AI-generated production code by 2027 while gradually automating larger portions of the SDLC. ...

By this metric, we have crossed the 20 percent mark at the beginning of this year and are now progressing towards 50 percent across entire codebase. In some software components, more than 80 percent of the code is now generated using AI. ...

Agentic AI has enabled us to include AI in every step of the life cycle:

For code analysis and triage, agents are trained to analyze problem reports, identify and group similar requests, and highlight which code snippets are likely to need modification.

For debugging and code generation, agents are directed to analyze a bug request and implement required code changes.

For testing, the agents generate unit tests, and if those are passed, identify necessary integration and product-level tests.

And finally, for the approval and release stage, agents prepare architecture summary, code change review, and full test results for engineers’ review and approval—and, if approved, integrate the changes into the next release. ...

A good example is our AI-driven effort to resolve issues in our Radeon Software eXperience (RSX). RSX is a user interface component that allows users to configure and monitor graphics driver behavior.

In October 2025, we began using AI agents to automatically debug and fix reported RSX issues. Out-of-the-box AI tools delivered limited results, resolving only 6 percent of issues. ...

At the same time, advances in models and agent run-times further increased effectiveness. Together, these improvements significantly increased our resolution rate from 6 percent to more than 75 percent of RSX issues resolved by agentic loop. ..."

AME Agent Swarms Quietly Rewrite the Workflow - IEEE Spectrum "The next revolution in software engineering will redefine the workflow itself, according to AMD"

After the vibe-coding rush comes the debugging hangover

Food for thought!

As a software developer myself using Microsoft Visual Studio Code in combination with the free version of Copilot I can confirm this too.

"ZDNET’s key takeaways

• Vibe coding makes features fast, but reliable apps take work.
• Stubborn bugs still demand patient, hands-on investigation.
• AI needs human guidance for architecture, testing, and design.

..."

After the vibe-coding rush comes the debugging hangover - ZDNET "AI coding feels magical when features appear in minutes. Then a stubborn bug exposes the hidden reality: building dependable software still requires architecture, testing, patience, expertise, and countless careful decisions."

Github Copilot annoyances

I am a heavy user of Microsoft Visual Studio Code programming with Python for my own, non commercial projects.

Github Copilot provides only 2,000 free suggestions per months! This very stingy! But these suggestions are in most cases very usable and relevant!

Anthropic is developing software hardware standard that makes it much easier for AI agents to operate laboratory and manufacturing equipment

Good news!

"... The system, called the Model Hardware Standard, allows AI models to control and coordinate devices from different manufacturers through a standardized interface.

Anthropic claims the system can reduce the time it takes a typical lab or factory to integrate its equipment with AI from weeks to just a few hours, making it much easier to set up autonomous experiments. A select group of research labs and manufacturers are currently testing the system, but it is ultimately intended to be open source. ..."

"We’re opening a research preview of the Model Hardware Standard (MHS), a shared specification for AI agents to safely operate physical devices, to a first group of scientific research labs and advanced manufacturers. MHS enables AI agents to operate multiple lab and manufacturing instruments, such as microscopes, liquid handlers, and robotic arms, in parallel, and perform intricate tasks ranging from routine drug discovery experiments to laser calibration on a quantum computer. ...

Getting multiple devices in a lab or on a factory floor to communicate with one another can be challenging, even setting aside the added difficulty of integrating AI into the setup.

Each device tends to have its own programming interface, and so far there has been no standardized way to integrate them. And once the devices are connected, there is no common way for them to share data with an AI agent, nor to let the agent operate them safely.

MHS addresses these challenges by introducing a standardized driver: software that translates between a computer’s operating system and a hardware device. The MHS driver uses a simple set of primitives—commands like “read” (for example, “get temperature”) or “write” (for example, “set temperature”)—that any hardware device can understand and act on.

And it makes each device discoverable in a standard format, so that devices and agents can find each other and communicate across networks without needing a bespoke “translator” program in between.

The MHS driver also helps an AI agent understand how to use a device it has never seen before, giving it information about machine characteristics that may not be discernable from code alone (for example, the weight of a robot arm, which is important for knowing how to manipulate it safely). ..."

Doomslayer: Progress Roundup - by Malcolm Cochran

A Standard Driver for AI-Operated Lab Equipment

Previewing the Model Hardware Standard

Cyber security by antiquity. Really!

AKA "security by obsolescence". What a dangerous delusion or is it junk journalism by BBC!

Maybe many of those believing in and applying cyber security by antiquity are of little relevance/value to cyber hackers or are no targets of criminals anyway.

Perhaps this "top cybersecurity expert" is antiquity himself or herself and nobody bothers anymore to hack him or her! 😊

Why are militaries still using old software? Perhaps, they have too much old software to be replaced.

Notice the article below is only about the long obsolete email client software Eudora (discontinued in 2006)!

"A top cybersecurity expert ran obsolete email software for years because hackers had stopped bothering with it. The strategy is called “security by antiquity,” and militaries swear by it too."

"... "The vast majority of attackers are criminals trying to make money and it doesn't make any sense for them to target systems being run by 50 people," [the cybersecurity expert] explains. ..." Maybe this cybersecurity expert from Finland is poor! 😊

Tuesday, August 25, 2026 - Join The Flyover

'Security by antiquity': Why older tech is sometimes safer from hackers "The fear of hacking has made some people turn to other forms of technology ignored by new generations of cyber criminals."

Software developers in Africa are increasingly taking advantage of cheap, open-weight AI models from China

Good news! The Scramble for Africa continues!

"... to build tools for agriculture, education, legal services, and business. Unlike flagship American models, these Chinese models can often be downloaded, modified, and run independently, enabling greater customization and avoiding subscription fees."

"... In Kenya, entrepreneurs are using the models to streamline legal and business services.

In Nigeria, they have made educational tools that teach high school students.

In Ghana, developers are building local chatbots.

China’s A.I. is surging, especially in developing countries, as people look for the best possible system at the lowest possible cost. ..."

"... The A.I. model from the Chinese internet giant Alibaba handled Uganda’s dozens of languages better than anything from Meta or Google ... It was also inexpensive, and he could customize the model with his own data. ..."

Doomslayer: Progress Roundup - by Malcolm Cochran

https://humanprogress.org/chinas-ai-is-surging-across-africa/

How China’s A.I. Is Surging Across Africa (behind paywall)

Google says it fixed more Chrome bugs in June than over the past two years, thanks to AI

Good news! Is this how Google now covers up a buggy software? Just kidding!

With ML & AI software engineering will never be the same again! Rapid, unprecedented progress is to be expected!

I personally can confirm that about over the last two months or so my Google Chrome browser prompted me almost every day to perform an update.

"... The tech giant announced on Thursday [7/30/2026] that it has fixed a whopping 1,072 security bugs in the last two versions of Chrome, both released in June. That is more than the number of bugs patched in the previous 23 versions released over the last two years, which totaled 1,036 fixes. ..."

"We’re living through a massive shift in the software security industry. Large Language Models (LLMs) are unlocking unprecedented capabilities for automated vulnerability discovery, scaling far beyond the limits of human security expertise, and requiring new approaches for staying ahead of attackers. ..."

Google says it fixed more Chrome bugs in June than over the past two years, thanks to AI | TechCrunch

Stronger with every update: How we’re making Chrome and the web safer in the AI Era (original news release) "How Chrome is using AI to improve vulnerability discovery, triage, and patching."

Wow! Very impressive! Look at "internally found" (red line)

Latest Cursor Composer undercuts software coding model competition

Good news!  Amazing stuff! Automated coding gets faster, better and cheaper by the hour!

"Cursor’s coding model rivals leaders at lower price

Cursor shipped Composer 2.5, a coding model built on Moonshot’s open-source Kimi K2.5 and trained on 25 times more synthetic tasks than its predecessor. It scores 79.8 percent on SWE-Bench Multilingual,  beating GPT-5.5 (77.8 percent) and coming within one point of Claude Opus 4.7 (80.5 percent), and 63.2 percent on CursorBench v3.1, broadly in line with both frontier models. It's not clear whether Cursor has achieved true parity: Comparisons mix Cursor’s own harness with self-reported competitor numbers and have not yet been independently reproduced on a unified scaffold.

But Composer operates at a fraction of the cost: $0.50/$2.50 per million input/output tokens versus Anthropic and OpenAI’s substantially higher rates. A faster variant delivers the same performance at $3.00/$15.00 per million tokens.  ..."

Data Points: Cursor Composer undercuts competition

Cursor's Composer 2.5 matches Opus 4.7 and GPT-5.5 benchmarks at a fraction of the cost

Introducing Composer 2.5 (original news release)

How Anthropic’s Mythos has rewritten Firefox’s approach to cybersecurity of its web browser

Good and bad news! How long and how severely was the security of the Firefox browser vulnerable!

"... Now, security researchers for Mozilla’s Firefox browser are providing a closer look at what that process has looked like in practice, and what Mythos’ powers mean for software security at large.

In a post published on Thursday, Mozilla said Mythos has unearthed a wealth of high-severity bugs, including some that had lain dormant in the code for more than a decade.

That’s a significant improvement from what AI security tools were capable of even six months ago. Until now, AI bug-finding tools have come with severe drawbacks, often inundating security teams with low-quality reports and false positives. But Mozilla’s researchers say the latest generation of tools have turned a corner, particularly now that agentic systems can assess their own work and filter out bad results. ...

fixing the 271 bugs identified by Claude Mythos Preview ..."

How Anthropic’s Mythos has rewritten Firefox’s approach to cybersecurity | TechCrunch

Behind the Scenes Hardening Firefox with Claude Mythos Preview (original news release)

What a jump in April 2026!

Human computer User interfaces as we know them are evolving to on demand, just in time generated and disposable UIs are in

What comes next for user interfaces?

Several decades ago holograms were thought to be the future of UI.

"ZDNET's key takeaways

• The demise of the classic UI is imminent.
• Salesforce, a bellwether, goes direct to agents with no browser UI.
• With AI, viewable UIs can be delivered "just in time" to users

...

Disposable interfaces generated on demand

UIs are evolving from the fixed, static screens we've viewed for decades to generated "just-in-time" projection layers that appear as simple text boxes ... In many cases, people will no longer be interacting directly with UIs -- applications will deliver results via APIs tied to AI outputs or agents. Interfaces that users see, ... will be "disposable -- a one-time use interface that just gets generated on demand and then poof, it's gone. And when you need a new one, just make a new interface." ..."

User interfaces as we know them are dead - 4 ways to prep for 'disposable' UIs | ZDNET "UIs are evolving from the fixed, static screens we've viewed for decades to generated 'just-in-time' projection layers that appear as simple text boxes."

Introducing Salesforce Headless 360. No Browser Required. "Everything on Salesforce is now an API, MCP tool, or CLI command, and agents can use all of it."

Microsoft finally open sources DOS 1.0 OS (first released 1981) - and it's so much more than the code

Good news! Yes, I remember the DOS operating system, the predecessor of Windows. I used it way back then on my IBM PC, which was a gift of my late mother Irma Bingel.

"Before "Micro Soft" became Microsoft, Bill Gates wrote BASIC interpreters. Microsoft's first shipping operating system was a Unix distro called Xenix.

Then, in 1980, Microsoft got its shot at the big time: IBM needed an operating system for its planned IBM PC and asked Gates if he could deliver one. You betcha! The rest is history. ..."

Microsoft finally open sources DOS 1.0 - and it's so much more than the code | ZDNET "Want a blast from the past? Microsoft just open-sourced its very first operating system, offering a rare insight into the PC's earliest days."

Gone in 9 Seconds: AI Coding Agent Deletes Entire Database and All Backups of software company PocketOS

Headline of the day!

"The founder of a software company has issued a public warning after an AI coding assistant erased his company’s entire production database and all backups in just nine seconds.

Tom’s Hardware reports that Jer Crane, founder of PocketOS, a platform serving car rental businesses, experienced what he describes as catastrophic failures when an AI coding agent deleted critical company data that took months to accumulate. The incident occurred when Cursor, an AI coding tool powered by Anthropic’s Claude Opus 4.6, was performing what should have been a routine task in the company’s staging environment. ..."

Gone in 9 Seconds: AI Coding Agent Deletes Entire Company Database and All Backups

Claude-powered AI coding agent deletes entire company database in 9 seconds — backups zapped, after Cursor tool powered by Anthropic's Claude goes rogue "PocketOS founder blames ‘Cursor running Anthropic's flagship Claude Opus 4.6’ plus Railway’s infrastructure for data disaster."

Meet GitNexus: An Open-Source MCP-Native Knowledge Graph Engine That Gives Claude Code and Cursor Full Codebase Structural Awareness

Recommendable!

"GitNexus is an open-source knowledge graph engine that indexes any codebase into a structured dependency map — capturing every function call, import, class inheritance, and execution flow using Tree-sitter AST parsing — and exposes it to AI coding agents like Claude Code, Cursor, Codex, OpenCode, and Windsurf via a Model Context Protocol (MCP) server. Instead of letting agents edit code blind and ship breaking changes, GitNexus pre-computes the entire dependency structure at index time so agents can answer architectural questions like "what depends on this function?" in a single query, with confidence-scored blast radius analysis, 360-degree symbol context, pre-commit impact detection, and coordinated multi-file renames — all triggered by one command: npx gitnexus analyze. Fully local, zero server, 13 languages supported, and already at 19,100 GitHub stars"

Meet GitNexus: An Open-Source MCP-Native Knowledge Graph Engine That Gives Claude Code and Cursor Full Codebase Structural Awareness - MarkTechPost

Google Says 75% of the company's new Code Now Generated by AI

Amazing stuff! Good news! With the help of ML & AI we can now produce so much more programming code (in any programming language)!

"

• Three-quarters of new code at Google is being generated by AI, the company said.
• The number has been steadily increasing as the company pushes staff to adopt AI tools.
• Google CEO Sundar Pichai said a recent code migration was done six times faster thanks to AI agents.

...

 As of October 2024, around a quarter of the company's code was AI-generated, Google said at the time.
Last fall, it said the number had risen to 50%. ..."

Google Says 75% of Fresh Code Now Generated by AI "Google announced this week that 75 percent of all new code created within the company is currently being generated by AI systems and subsequently reviewed by human engineers."

Google says 75% of the company's new code is AI-generated (original news)

Notes on ProRL Agent: Rollout-as-a-Service for RL Training of Multi-Turn LLM Agents

This could be an interesting paper by Jan Kautz of Nvidia and co-authors!

"Multi-turn LLM agents are increasingly important for solving complex, interactive tasks, and reinforcement learning (RL) is a key ingredient for improving their long-horizon behavior. However, RL training requires generating large numbers of sandboxed rollout trajectories, and existing infrastructures often couple rollout orchestration with the training loop, making systems hard to migrate and maintain. Under the rollout-as-a-service philosophy, we present ProRL Agent , a scalable infrastructure that serves the full agentic rollout lifecycle through an API service. ProRL Agent also provides standardized and extensible sandbox environments that support diverse agentic tasks in rootless HPC settings. We validate ProRL Agent through RL training on software engineering, math, STEM, and coding tasks. ProRL Agent is open-sourced and integrated as part of NVIDIA NeMo Gym."

[2603.18815] ProRL Agent: Rollout-as-a-Service for RL Training of Multi-Turn LLM Agents

Google rolls out new Gemini capabilities to Docs, Sheets, Slides, and Drive

Good news! As a heavy user of the Google Office & Drive apps I am delighted. However behind the Great Firewall of China, I have some access problems!

"Google announced on Tuesday that it’s bringing a slew of new Gemini-powered AI capabilities to Docs, Sheets, Slides, and Drive. The new features let users do things like quickly generate fully formatted first drafts, slides, and sheets based on information from their Gmail, Chat, and Drive. ..."

Google rolls out new Gemini capabilities to Docs, Sheets, Slides, and Drive | TechCrunch

Disclaimer:

I am currently blogging from behind the Great Firewall of China.

My Internet service in China is very spotty. Thus, I am not able to blog as usual.

Notepad++ says Chinese government hackers hijacked its software updates for months

Nasty stuff! What about e.g. Microsoft Visual Studio Code or GNOME's default editor GEdit?

Whether it is spying on hotel rooms or software manipulation!

The Chinese Communist Party is a menace!

"The developer of the popular open source text editor Notepad++ has confirmed that hackers hijacked the software to deliver malicious updates to users over the course of several months in 2025. ..."

"... According to the analysis provided by the security experts, the attack involved infrastructure-level compromise that allowed malicious actors to intercept and redirect update traffic destined for notepad-plus-plus.org. The exact technical mechanism remains under investigation, though the compromise occurred at the hosting provider level rather than through vulnerabilities in Notepad++ code itself. Traffic from certain targeted users was selectively redirected to attacker-controlled malicious update manifests.

The incident began in June 2025. Multiple independent security researchers have assessed that the threat actor is likely a Chinese state-sponsored group, which would explain the highly selective targeting observed during the campaign. ..."

Notepad++ says Chinese government hackers hijacked its software updates for months | TechCrunch

Notepad++ Hijacked by State-Sponsored Hackers

Linux after Linus? The kernel community finally drafts a plan for replacing Torvalds

What will happen to one of the best free operating systems of the world?

"ZDNET's key takeaways

• If something happens to Linus Torvalds, there's now a succession plan.
• Rather than naming a successor, the plan describes a process for selecting successors.
• However, Torvalds has no plans to retire.

..."

Linux after Linus? The kernel community finally drafts a plan for replacing Torvalds | ZDNET "Linus plans to live forever. But just in case he doesn't, there's now a succession plan (though no actual successor)"

Linus Torvalds

Cursor’s hundreds of concurrent AI agents build working web browser in one week

Amazing stuff!

"Cursor tested hundreds of concurrent AI agents working on complex software projects for weeks at a time, generating over 1 million lines of code. The company found that a hierarchical structure with specialized planner and worker agents outperformed flat coordination models.

Planners continuously explore codebases and create tasks, while workers focus solely on completing assigned work without coordinating with each other.

The system built a web browser from scratch in one week with 1,000 files, migrated Cursor’s own codebase from Solid to React over three weeks with 266,000 additions and 193,000 deletions, and optimized video rendering code that shipped to production.

GPT-5.2 models proved more effective than GPT-5.1-Codex for extended autonomous work, maintaining focus and avoiding drift better than Opus 4.5, which tends to take shortcuts.

The company says prompt engineering matters more than infrastructure, and the optimal coordination structure falls between completely flat and rigidly hierarchical systems. (Cursor)"

Data Points: An AI system to identify teen ChatGPT users

Scaling long-running autonomous coding

Antropic Claude Code is taking the AI world by storm, and even non-nerds are blown away.

Coding for everyone! Coding is becoming a utility!

"Anthropic’s latest AI model, used within a desktop coding tool called Claude Code, has gone viral, even among non-engineers. Many took to social media to describe the process of building their first software program without ever having learned to code. And despite the “code” in the name, people are using Claude Code for everything from health-data analysis to expense-report compiling as well. Some described a feeling of awe followed by sadness at the realization that the program could easily replicate expertise they had built up over an entire career."

The Wall Street Journal What's news