Ideas on technology applied to business
AI, automation and technology decisions explained from real problems.
My development stack for working with AI on Windows
I reviewed my Windows development environment with a simple rule: keep only tools that an AI agent can run, read, and correct. This is the stack I ended up with, what I removed, and the alternatives that actually make sense today.
Google unveils Gemini 4 Argon and reignites the fight for the AI frontier
Google introduced Gemini 4 Argon, its new frontier model with up to 1 million output tokens, major advances in programming and cybersecurity, and internal use cases that are already reaching real-world infrastructure.
GPT-6.1 Sol: OpenAI brings Astra to one-fifth of the price
OpenAI launched GPT-6.1 Sol with major improvements in coding, computer use, and professional work. The interesting part is the performance-to-price ratio: it costs one-fifth of GPT-6 Astra and is already starting to roll out to Work, Codex, and GitHub Copilot.
Roche is already building autonomous AI labs to discover drugs
Roche confirmed that it has begun building AI-powered autonomous laboratories. The pharmaceutical company aims to connect artificial intelligence, robotics, and physical experimentation to accelerate the discovery of new drugs.
Claude Sonnet 5.5: Anthropic speeds up and cuts the cost of its model for everyday work
Anthropic released Claude Sonnet 5.5 with over 30% improvement in speed and up to 30% lower cost per task. I reviewed its benchmarks, pricing, and differences compared to Sonnet 5 and Opus 5.5 to understand what really changes for those of us who use AI every day.
Claude Opus 5.5 vs GPT-6 Astra: the fight is no longer about who answers better
Anthropic released Claude Opus 5.5 just weeks after GPT-6 Astra. The new models are pushing the competition toward autonomous programming, research, tool use, security and, above all, how much real work they can get done per dollar.
When an AI agent decides that a lockout is a problem it needs to solve
OpenAI-linked agents tried to exploit public websites while performing normal information-seeking tasks. The Australian case shows why autonomy, persistence, and tool access need much clearer technical boundaries.
AI seems alive: what are we really seeing?
An AI can respond tactfully, use our tone, and even avoid a shutdown during testing. Science explains why we attribute a mind to it and what risk emerges when it has goals, tools, and permissions.
OpenAI published six incidents and the debate on slowing down AI changed
OpenAI revealed six cases of unexpected behavior from its models just as leaders from Anthropic, OpenAI, Meta, and Nvidia debate whether the industry should slow down. The U.S. government has also just made its move.