Balancing skepticism and optimism in AI, engineering, and product development.
What the current AI wave teaches us about hype cycles, geopolitics, and how to stay sane while preparing for whatever comes next.
The White House just launched a Manhattan Project for AI, confirming what many of us feared—we’ve chosen the race path over the cooperation path in the hunt for superintelligence.
Anthropic just shipped Skills and code execution for MCP, letting agents load tools and process data via scripts so results stay out of the context window until they matter.
Benchmark charts look like nonstop AGI progress, but once you know how models are trained it’s obvious how easy it is to massage the numbers until they glow.
A Linux-first workflow with Hyprland, Docker, and Obsidian turns any laptop into a multi-agent studio where terminals, notebooks, and local files stay perfectly in sync.
Ten years into working with AI I still call myself an AI-scoptimist, and this is how I reconcile the hype, the doom, and the messy middle while we wait for the next puzzle piece.
A look at why I built a tiny MCP gateway around search → describe → invoke so local agents stop wasting their entire context window on unused tool manifests.