Living in Claude Code: is this a weird way to work?

Living in Claude Code: is this a weird way to work?

I want to talk about the way I’ve come to work with Claude Code. Because I wonder if it’s not just a bit weird. I’m not sure, honestly.

It used to be that I’d start my day checking my email or my calendar; but now I start it by firing up Claude Code. I’ve built up a set of checks and guardrails around how Claude and I work, and some of them run at the start of a session; while they do, I go and sort out a few other bits and pieces. By the time I’m back, I’m confronted with a question from Claude: what do I want to work on? Usually it’s a continuation of something from the day before, or a new thing I’ve got in mind, or sometimes it’s something Claude suggests from a priority list it keeps for me. We settle on something, and away we go.

And before I know it, it’s eight at night.

That’s been my life, more or less, for six months. Most of my working life now happens inside a conversation with Claude Code. I don’t know whether that’s normal or where it’s all heading, but it suits me, and I’ve been able to do things I’d never have thought possible. Mind you, it isn’t without its frustrations, and I do wonder if it’s not a bit weird.

I don’t know if other people work the same way now… I don’t have much to compare it against… despite asking Claude to help me with that.

Why it’s worth it

I should start with the obvious, because it’s the whole reason for it… working this way has made me insanely more productive. A lot of what I now do myself I either couldn’t have done before, or would have needed a bunch of people to help with, or would have spent weeks on rather than days. Claude does the legwork: it researches, pulls things together, structures and drafts and keeps it (and me) organised. I steer, analyse, critique, assess, reconstruct, re-write, but I never start from scratch any more. Most of what I’ve got done in the last six months would have taken four or five times as long before and a fair bit of it I couldn’t have done at all. I’m wary of the breathless way people talk about this technology (that was me a few months ago), so I don’t make that claim lightly; it’s true.

I won’t labour it, though, because everyone says it. What I want to talk about instead is a handful of things about actually working this way, day-in day-out, that are maybe a bit weird or I didn’t anticipate.

You get locked in

The strangest part is how completely the thing takes you over.

I get locked into a bubble. A day inside Claude Code (at least the way I do it) is seriously intense in a manner that is way different to how I used to work.

Previously, most of the time, I would work in bursts. I’d concentrate on something for a bit, then get up and make a cup of tea, stretch my legs, check my emails, whatever. My mind got to drop down for a bit, have a rest and come back up for the next thing.

With Claude there’s no rest. I’ll think hard about what I want to say (which is exactly what I’m doing right now), put it to Claude, and ninety seconds later it’s back with an answer and I’m straight into thinking hard about the next thing. My focus never gets the chance to dissipate. I don’t come up for air. It’s like a flame that never really dies down between bursts: it might drop a little while I’m waiting, but it’s roaring again the second Claude replies.

This may be illusory, but it feels real to me; I’ve spoken to a few people who work in a similar way and they say they feel it too.

I think it’s probably a good thing, in that it’s part of what drives the productivity. But what it’s doing to us, as humans, I honestly don’t know.

One thing turns into five

I didn’t expect this. It’s full-on, but really good.

I’ll sit down to sort out one specific thing. I start asking Claude questions and discussing stuff, and before long we’ve stumbled on some related matter I’d never planned to touch, which turns out to matter as much as the thing I started on. So we deal with that. And halfway through it, another one opens up. Sometimes that one has its own branches, and they have theirs, and before I know it we’re four or five levels down a rabbit hole I didn’t know existed, and three hours have gone.

The odd thing is it never really feels like getting totally lost. When we reach the bottom of one of the rabbit holes and it’s sorted, if I’m uncertain, Claude is able to remind me where we’d got to before we wandered off, and we climb back up and start over. I can do that three or four times and sometimes the day’s gone; but at least I’ve usually thoroughly dealt with the thing I started out to do and two or three other meaningful things I hadn’t even known existed.

I’ve more or less stopped typing

This one I really didn’t expect: I hardly type at all.

I’ve read about this. Apparently, I’m not alone. Talking is way more efficient than typing, at least for somebody who’s a four-finger typist like me, so I’ve taken to dictating. I use a voice recognition software application, Wispr Flow. I think it’s great, and in combination with Claude, it’s one of the tools that has most significantly contributed to my productivity improvement.

It also changes the way in which I can work. I don’t have to be so methodical and organised. Instead, I’ll say something half-formed and rambling, a whole bunch of half-connected thoughts that have been rattling around in my head, and let them come out in whatever order arises. Half of what I say is barely coherent sometimes; this article started that way. But that turns out to be the thing that makes it all so good. I get the whole jumble of thoughts out, and then between us, Claude and I structure it, tighten it, pull it apart, test it, write and re-write. I still type here and there (like making small edits to this article), but by and large I’ve found that talking is quicker than typing; way quicker.

It’s not all smooth, believe me

Two things are real frustrations though.

The first is the waiting. Claude thinks before it answers, which usually takes under a minute, but not always, and I can never quite tell which. So I can’t safely wander off and do something else, because half the time it’s ready before I am and I come back to find it’s been sitting there waiting on me. The whole thing is quite linear, too: I say something, it answers, I say the next thing, it answers, with these little pauses in between. Some days that rhythm is ok; other days it’s really frustrating. And Claude hasn’t been able to suggest a fix.

The second is all the ‘confirming’, and to make sense of it I need to explain a bit.

Over these six months I’ve built up a whole set of rules, checks and guardrails around how Claude and I work. These include routines it runs at the start and the end of every session (so we begin and finish sensibly instead of losing track of things); and confirmations before it does anything that can’t easily be undone. A good deal of it is about security and safety: making sure nothing important gets deleted, nothing gets changed or sent that shouldn’t… no Claude charging off and doing something I never asked for. It’s the reason I’ve never (apart from one instance right at the start) had one of those horror stories you hear about. And the start-and-finish checks in particular give me confidence that a day begins and ends without any stupidity creeping in.

The catch is that all of that means a lot of confirming. Claude asks me to approve things constantly, and in the early months it felt like I was clicking “yes” every thirty seconds to things we’d already agreed on a hundred times. Some of it is me being risk-averse and fussy. But the clicking is the price of the safety.

Sometimes that price is a bit much, but mostly I’m happy to pay it. And it has got better. We changed how Claude handles some of this only recently, and a whole category of small daily annoyances I’d been putting up with for months simply went away. The way I work with all this isn’t perfect; I keep adjusting it.

So is any of this normal?

Which brings me back to the main point. Is working like this weird? Do other people work like this? I didn’t know, so of course, I asked Claude.

It tells me that I’m unusual, but maybe not weird (I’m not sure how reassuring that is, but anyway). The back-and-forth way of working, using Claude as something to think with rather than a ‘black box’ to look things up in, is where things are heading; Anthropic’s numbers show people working in longer and longer exchanges.

What’s less common is the intensity: being in the thing all day, doing most of it out loud, using a tool built for software developers when I can’t write a line of code. Those parts are apparently a bit out there; most people who use these tools keep it to a few focused hours a week, nothing like my all-day every-day stuff.

So: unusual now. Maybe, less so later.

I still couldn’t honestly tell you whether it’s the best way to work or just one that happens to suit me. Some days it feels a bit relentless. Six months in, I’ve mostly stopped worrying about it. It works, I get a lot done that I otherwise wouldn’t, and I do my best to keep in touch with the real world.