Previously at the Volcano Base I’d been waxing (sic) lyrical about lubrication. Since then, honestly, I’ve been waiting for a client to give me a password. While waiting, I’ve chipped-in on some new business opportunities, and used Claude Code and Codex to create a cashflow forecasting application that I’d previously been paying a SaaS subscription for. The benefit of making your own software is that you can customise it to just do the things you need it to do.

New tools have a remarkable talent for making everything else look broken. A workflow that worked perfectly well yesterday is suddenly “unsustainable”. A spreadsheet becomes a risk exposure. A demo video assures you this is all very serious.

The “do nothing” benchmark is the unfashionable pause before you fix something, where you ask a dangerous question: what if we didn’t?

This isn’t about neglect or stubborn nostalgia. It’s about recognising that the current system, even if it’s scruffy and mildly irritating, is a valid option. Leave it alone. Observe it. See what actually happens.

Most of the time, nothing dramatic occurs. The report still goes out. Clients still pay. Tuesday continues to arrive on schedule.

That baseline matters, because every improvement carries luggage. Some of it is obvious: setup time, subscriptions, and the brief period where everyone forgets how buttons work. The rest is sneakier. Mental overhead. Fragile integrations. Time spent building something clever that saves almost no time at all.

Compared to doing nothing, these costs stop being theoretical. The “quick win” reveals itself as a small expedition through documentation and polite frustration.

This isn’t anti-technology. It’s anti-faff. The businesses that endure aren’t the most elaborate. They fix the problems that genuinely hurt and leave everything else comfortably alone.

Next time a new system is suggested, write down exactly what happens if you do nothing. Be specific. Be honest. Then notice which option makes you tired just thinking about it.

Sometimes the most strategic decision is to change absolutely nothing.

Some interesting stuff I discovered on my adventures.

AI agents go shopping without you

Google has announced a new open protocol designed to let AI agents handle commerce end-to-end, from product discovery to checkout and support.  

Why it matters: If this works, e-commerce becomes something that happens quietly in the background while humans argue about something else entirely.

A small, open robot you can actually play with  

Reachy Mini is an open-source desktop robot aimed at developers, educators, and the merely curious who want to experiment with embodied AI without a research lab.  

Why it matters: When robots become affordable and hackable, experimentation gets weird in a good way.

A constitution for AI  

Anthropic’s constitution lays out the principles guiding how its models should behave, including safety, transparency, and alignment with human values. Having read this a long time ago, revisiting it this week reinforced its importance for me.

Why it matters: It’s a rare attempt to write ethical intent down before everything gets very complicated.