Philippos Savvides

updates

I built an interactive Monte Carlo simulation of the US-Iran war. 100 nodes. 8 phases. 5,000 scenarios per run. Bayesian network with real data from Feb 28 onward: savvides.github.io/usairanmonteca… Open source: github.com/savvides/usair…

I published my 2018 PhD dissertation on GitHub with reanalyzed the data. The core finding held up (d > 2.4). Open data, updated lit review, and an honest retrospective github.com/savvides/disse…

Introducing the Google Workspace CLI: github.com/googleworkspac… - built for humans and agents. Google Drive, Gmail, Calendar, and every Workspace API. 40+ agent skills included.

1999: Small, lean, quick, fit, profitable. 2026: Small, lean, quick, fit, profitable. The fundamentals are the fundamentals.

Probably the most important question founders need to be asking every day: how do we survive commoditization?

This is an excellent thought exercise. In reality physics, fear, legal liability, and government self-preservation would aggressively slam the brakes on this specific timeline.

3 months ago, I quit my job to chase a dream: to build an affordable, convenient, plastic-free coffee maker. Grateful for everyone who has reserved ❤️ puresteelco.com

“It shall be called the Super Bowl and will be a huge national event played deep into the night on Sunday when everyone must go into work on Monday.” “Why not play it on Saturday?” “Nobody knows.”

4% of GitHub public commits are being authored by Claude Code right now. At the current trajectory, we believe that Claude Code will be 20%+ of all daily commits by the end of 2026. While you blinked, AI consumed all of software development. Read more 👇 newsletter.semianalysis.com/p/claude-code-…

A simple heuristic for where some of the biggest AI agent opportunities are: domains where reading or writing lots of information is a core bottleneck of value creation. Coding has been the first and most obvious one so far where this manifests in a big way. But the same will hold for areas constrained by reading or writing contracts, financial documents, emails, medical information, logs, and so on. There are probably dozens or hundreds of categories that still haven’t been addressed yet of what happens when you lower the cost of reviewing or producing any form of data by 99%.

Things ChatGPT told a mentally ill man before he murdered his mother:

@alex not surprised. Any details on the deal ?

Woah what manus.im/blog/manus-joi…

Real Luxuries in Life 1. Living 10 minutes from work 2. Living 5 minutes from the gym 3. Having quiet neighbors 4. Having money left at the end of the month and investing it 5. Peace at home 6. Drinking coffee without rushing 7. Sleeping with a clear conscience 8. Laughing with people who truly get you 9. Traveling every year 10. Waking up naturally without an alarm 11. Enjoying a home-cooked meal with loved ones 12. Having time to read a book in one sitting 13. Finding joy in simple daily routines 14. Having a pet that greets you happily at the door These are the things that actually feel rich.

Not sure many fully grasp what's coming.

3,000 years in the making youtu.be/Mzw2ttJD2qQ?si… via @YouTube

Need a projects UX @GeminiApp

Finally: fitness is not static. If you are training hard, your physiology changes fast—especially in the first few months. Don't let your data expire. Re-test your FTP once a month to capture your gains. Training at old numbers = Training in the past.

Crucial Step: Use 6 Zones, not 5. Standard 5-zone models blur the line between VO₂ Max and Sprinting. Set Zone 6 at >121% FTP. If you hit Zone 6 during a 4x4, you are going too hard. Cap your effort in Zone 5 (106–120%).

I wasted months of training because I trusted an algorithm over physics. I relied on my Apple Watch to estimate my fitness. I treated the data as gospel. I was wrong. Here is how I fixed a 19% error in my baseline and why you need to stop training by Heart Rate. 🧵👇