The Day Claude Went Down (and the World Kept Turning)

So Claude fell over today.

For a couple of hours there, you could practically hear the collective sigh across every dev team, marketing department and "I'll just get AI to draft it" inbox in the country. The error message popped up and, honestly, the reactions were a bit dramatic:

• "Claude's down. Right, that's me, pack it up, see you tomorrow."
• "What are we supposed to do now, type our own emails like animals?"
• "Quick, does anyone remember how to think?"

I'll admit I had a little chuckle, mostly because I'd caught myself doing exactly the same thing. Tab open, prompt typed, hit enter... spinner... nothing. And then that brief, existential pause where you realise how much of your day quietly runs through a chatbot now.

We got a bit too comfortable, didn't we

Let's be honest with ourselves. AI tools are brilliant, and I use them every day. But somewhere along the way a lot of us went from "this is a handy assistant" to "I cannot begin this task until the machine wakes up." A two-hour outage shouldn't bring a whole team to a standstill, and yet for some places today, it basically did.

It's a bit like when the office coffee machine breaks. Technically you can still do your job. You just suddenly feel like you can't, and everyone stands around looking wounded.

How to not down tools when the robots do

Outages happen. Every tool you rely on - AI, your CMS, your email, your hosting - will go down at some point. The trick isn't avoiding it, it's not being completely useless when it does. A few genuinely useful habits:

1. Keep a "no-AI" task list. There's always work that doesn't need a model: reviewing pull requests, writing tests, tidying the backlog, that documentation everyone's been avoiding. When the AI's out, that's what you pull from.
2. Don't let one tool be a single point of failure. If your entire workflow collapses because one service hiccups, that's a process problem, not an outage problem. Have a fallback, even a clunky one.
3. Actually keep the skills sharp. If nobody on the team can write a function, a proposal or a decent email without AI holding the pen, that's worth fixing quietly, before an outage fixes it for you.
4. Check the status page before the group chat melts down. Nine times out of ten it's a known issue with an ETA, and you can just... go and do something else for an hour.
5. Treat downtime as thinking time. Some of the best decisions get made when you're forced away from the screen. Go for a walk. Sketch the thing on paper. Talk to an actual human colleague.

The slightly more serious bit

There's a real lesson buried under the memes, and it's the same one I bang on about with websites: don't build your whole operation on something you don't control and can't fall back on. AI is a fantastic multiplier when it's working. It shouldn't be the load-bearing wall.

The teams that barely noticed today weren't the ones with no AI. They were the ones who treat it as one tool among many - useful, not essential - and had enough of their own process, skills and common sense to keep moving without it.

Anyway, Claude's back now. We can all stop pretending we were about to write that report by hand.

If your business is a bit too reliant on one wobbly thing - a tool, a plugin, a mystery website nobody can log into - that's exactly the sort of thing I help untangle. Give me a shout.