What to Do While I Wait for Claude

I’m writing this in the 45-second void that defines my work life these days – the agonizing little eternity while I wait for Claude to respond. Yes, that Claude, the oh-so-clever AI assistant that’s become my daily co-builder and occasional bane. I spend all day building incredible things with Claude, but I also spend an incredible amount of time watching a blank screen, tapping my foot, and questioning my life choices as I wait for it to do its thing. If you’ve ever found yourself brewing yet another cup of coffee or doomscrolling through news feeds while your AI is “thinking,” pull up a chair. We have much to commiserate about. You and I. that Claude do its thing Don’t get me wrong: Claude is brilliant – chef’s kiss levels of brilliance when it works. It can debug code, brainstorm architecture, even refactor that one legacy method nobody on the team understands. It’s the ultimate pair-programming buddy… until it isn’t. Until I ask it a complex question and it goes into deep meditation mode, leaving me staring at the spinning cursor like it’s an oracle conjuring the secrets of the universe. I know there are real technical reasons for the delays (big model + big tasks = big waits, etc.), but in the moment it just feels like my AI sidekick decided to take a coffee break right when I needed it most. And here I am, a supposedly cutting-edge developer at a big tech company, reduced to waiting. In a world of instant AI magic, I’ve discovered a new form of torture: hurry up and wait. chef’s kiss real waiting hurry up and wait The Claude Conundrum: Hurry Up and Wait Working with Claude is a tech worker’s dream and nightmare rolled into one. On one hand, it’s like having an on-demand genius who can turn “hours of frustrating investigation into minutes of productive problem-solving”(anthropic.com). At least that’s the marketing promise. The reality? Those “minutes of productive problem-solving” often come after several minutes of watching Claude’s little loading indicator and wondering if the model fell asleep. It’s as if I have a supercar in the garage, but every time I hit the gas, I get stuck in LA traffic. The potential is awe-inspiring; the latency is soul-inspiring – as in, it inspires my soul to leave my body while I wait. anthropic.com several latency The tension between speed and delay has turned me into a parody of a Silicon Valley cliché. By day, I’m the enthusiastic AI evangelist boasting how “Claude helped me ship this feature in 2 hours flat.” By late afternoon, I’m the exhausted skeptic muttering that I could’ve Google-searched the docs faster than waiting for Claude to generate the same answer. (In fact, one old-school dev on Hacker News literally argued that Google searches beat “typing out your problem and waiting for ChatGPT to generate results”(news.ycombinator.com). On cranky days, I feel that.) It’s a bizarre ping-pong of productivity: Claude accelerates my work… except when it brings it to a dead halt. And I wait and wait and wait. literally news.ycombinator.com And of course, it always feels like it slows down right when I’m “in the zone.” You know, that precious flow state where your fingers fly and your brain is piecing together code like a beautiful mind. Then wham! – Claude needs a minute (or five) to process. My flow shatters. I go from feeling like Tony Stark with an AI JARVIS, to feeling like a frustrated kid watching a progress bar inch along. It’s the Claude Conundrum: I can’t live without it, but with it I’ve had to relearn the ancient art of patience. right when I’m “in the zone.” wham! with AI Hype vs. AI Reality: Living on the Bleeding Edge (of My Nerves) If you read the tech headlines or listen to our CEO at all-hands, you’d think AI has turned work into a utopian playground of effortless productivity. “AI will 10x your output! It’s a new era of efficiency!” they proclaim. And sure, some days it feels like that – like I have a genius intern who never sleeps. But other days, I’m painfully aware of the gap between AI hype and AI reality. AI culture right now is intense. My colleagues swap prompt-engineering war stories over lunch like old fishermen bragging about big catches (“I got Claude to optimize our entire database query with one prompt!” ... “Oh yeah? I got it to write unit tests in the style of Dr. Seuss!” Actually, no one brags about that but I do like to visit the Dr. Seuss originals by Fisherman’s Wharf). It’s equal parts exhilarating and exhausting. Everyone’s trying to prove they’re surfing at the crest of the AI wave, automating everything, riding into the transhuman sunset. “AI will 10x your output! It’s a new era of efficiency!” AI culture right now is intense. Meanwhile, here I am, literally staring at a screen, waiting for an AI to respond. It’s the ultimate reality check. AI may be transforming work, but it hasn’t eliminated those mundane moments of downtime. In fact, it’s introduced a weird new species of downtime: waiting for the machine to catch up with you. I sometimes picture a support group for high-powered developers sheepishly admitting, “Hi, my name is Alex and sometimes I lose productivity because I’m waiting on AI.” The horror! transforming machine you AI The culture around me doesn’t exactly help. We have this ethos in tech of move fast, be always on, crush productivity blockers. Waiting for your tools was not supposed to be one of those blockers in 2025. It almost feels taboo to acknowledge. Complaining that “Claude is slow today” in the team Slack is like sacrilege when everyone else is busy gushing about how AI is their superpower. But I know I’m not alone. Beneath the hype, there’s a quiet undercurrent of folks experiencing the same thing – the little hiccups and delays, the times when the AI isn’t instant and you have to find ways to cope. We’re living on the bleeding edge, and sometimes that edge cuts into our schedule. move fast be always on crush productivity blockers So yes, I love AI. I love Claude like a colleague (one that I suspect might unionize for more break time given how often it takes a pause). But I also find myself in absurd situations due to this love. Like rage-scrolling through Hacker News threads about “AI will replace programmers” while I wait for AI to, you know, help me program. The irony is richer than a Google cafeteria chocolate mousse. AI culture says I’m supposed to be turbo-charged; the reality is I’m sometimes twiddling my thumbs, performing what I call productiveness theater to feel like I’m still a good little worker bee. love Hacker News while twiddling my thumbs productiveness theater Productiveness Theater: Performing Busy-ness (Because Actual Work is on Pause) Ah, productiveness theater – I’ve come to know it well. In a nutshell, it’s when you perform productivity to seem (or feel) busy, rather than actually being meaningfully productive. It turns out, waiting for Claude is the perfect time to put on this one-person show. Instead of admitting “I can’t do much until Claude finishes its output,” I instinctively start doing something, anything that looks like work. According to one explainer, productivity theater is all about “prioritizing tasks that make [you] seem more productive, more available and busier than [you] really are”(worklife.news). Ding ding! That’s me in a nutshell, at least whenever Claude’s latency strikes. productiveness theater perform something anything worklife.news Suddenly I’m replying to emails within 30 seconds, chiming in on Slack threads, updating JIRA tickets at lightning speed – embellishing responsiveness for all it’s worth. If you walked by my desk during one of Claude’s “thinking” cycles, you’d see a flurry of alt-tabs and the furious clicking of someone who must be accomplishing a lot. In reality, I’m killing time and overcompensating. I might even fire off a few “Just circling back on this!” messages to team members (I am so so good about circling back), because hey, gotta look busy. It’s basically the remote-work version of leaving an empty Excel sheet open when your boss walks by. embellishing responsiveness must I’ll confess to some truly absurd behaviors in the name of productivity theater. I’ve caught myself nodding earnestly on a Zoom call I wasn’t paying attention to, purely to look engaged, while my mind was actually fixated on whether Claude had finished running a code analysis in another window. I’ve also mastered the art of the fake deadline face: furrowed brow, intense stare at the screen, which I deploy if a manager pops into the video chat – even if at that moment I’m actually reading a Wikipedia article about the Succession series finale (hey, I needed something to do while waiting). Anything to signal productivity. nodding earnestly fake deadline face Succession signal It’s not just me. Apparently this is a whole phenomenon now. People run videos of themselves or use mouse-jiggler apps to simulate activity when they’re actually (AFKworklife.news). While I haven’t gone that far, I have created the illusion of coding when I was actually just tinkering with a side project unrelated to work. In fact, I recently read an anecdote of a bored employee who would even switch their VS Code editor to a light theme so it looked more like “normal” work apps, and then write code to procrastinate on the job – “instead of using social networks like normal people do”(news.ycombinator.com). The dedication! By comparison, my trick of feverishly reorganizing my browser bookmarks while waiting for Claude seems downright tame. worklife.news have write code to procrastinate news.ycombinator.com Let’s be clear: productiveness theater isn’t real productivity. It’s cosmetic. It’s me trying to reassure myself (and anyone watching) that I’m useful and active, even when I’m temporarily at the mercy of a lagging AI. It’s a bit self-aware, a bit tongue-in-cheek – I know I’m play-acting. But it beats just sitting there feeling stuck. And who knows, maybe some of these little tasks have marginal utility (my inbox was zero for like 12 glorious minutes yesterday!). In the grander scheme, though, I’m basically performing “busy” to fill the void. It’s like being an actor in a one-man show called “The Developer Who Definitely Isn’t Just Waiting on a Machine Right Now.” 👀 real zero “The Developer Who Definitely Isn’t Just Waiting on a Machine Right Now.” 11 Ways to Stay Productive(ish) While Claude is “Thinking” So, what do I actually do while waiting for Claude? How do I attempt to “stay in flow” during those limbo moments? Short answer: with great creativity and even greater self-delusion. I’ve tried about every coping mechanism in the book (and a few that probably deserve their own DSM entry). Here, in no particular order, are 13 ways I pretend to be productive during Claude’s longest pauses – roughly 13 flavors of productive procrastination: actually do productive procrastination The Tab Tornado – I open, oh, about 27 new browser tabs on a whim. Documentation I’ve been meaning to read, that interesting blog someone posted in Slack, random Stack Overflow questions tangentially related to my work. My Chrome looks like it’s cosplaying as my brain: a million things open at once. Naturally, I’ll only actually read 2 or 3 of these tabs (and forget about the rest), but for that brief moment, I feel busy. Pro tip: more tabs = more productivity, right? (No. The answer is no. But in the moment, it sure feels like maybe.) Micro-Side-Project Mania – In a fit of “I’m totally using this wait time constructively,” I sometimes spin up a tiny side project. This could be as trivial as writing a quick script to automate a personal task, or as over-ambitious as saying “I’ll just start building that app idea I had.” I’ve literally generated a new React app and begun tinkering while waiting for Claude. By the time Claude returns with my answer, I’ve forgotten what I was doing in the main project because I’m knee-deep in a completely unrelated coding escapade. It’s the developer equivalent of stress-baking a cake while procrastinating on writing a report. The Snack Expedition – If all else fails, wander to the kitchen. I can justify this because technically I’m optimizing my personal fuel levels, right? I’ll grab a coffee, then think, maybe I need a snack. Next thing I know, I’ve crafted a seven-layer peanut butter and Nutella sandwich (don’t ask) or eaten half a bag of pretzels while staring into space. I’ve had days where my caffeine intake correlated disturbingly with Claude’s response times – the slower the model, the more espresso shots consumed. The outcome: a very jittery developer who has answers from Claude but also hands shaking too much to type properly. VS Code Theme Makeover – This is a classic “productive” procrastination move: tweaking my development environment. Did I need to download three new Visual Studio Code themes and A/B test which one gives me the best vibe? Of course not. Did I do exactly that while Claude was churning on a particularly long response? Absolutely. There is a special kind of nerdy joy in telling yourself that changing your editor’s color scheme to Nordic Ice or Neon Vamp will somehow make you more productive. It won’t, but it will beautifully consume 10 minutes. Plus, if anyone glances at my screen, they’ll think I’m just super detail-oriented about my tooling (instead of simply desperate for any task to focus on). Inbox Zero Heroics – Normally, I treat my email inbox like radioactive waste (look but don’t touch unless absolutely necessary). But in moments of AI-induced downtime, suddenly I’m an email-clearing machine. Newsletter I never read? Unsubscribe! Some auto-generated alert? Archive! I even start writing overly prompt replies: “Apologies for the delay, see my answers inline…” – as if the 4 minutes it sat unanswered was a grand inconvenience to the sender. The irony: I’m only ultra-responsive because I’m waiting on Claude. If my colleagues knew my sudden email diligence was a side effect of AI slowness, they’d chuckle. But hey, at least for a moment I feel on top of my correspondence – a veritable Inbox Zero Hero, cape and all. Rubber Duck Debugging (Literally) – This one’s borderline embarrassing, but here we are. I have a rubber duck on my desk (a classic programmer trope) for debugging by explaining problems out loud. Sometimes, while Claude is cooking up an answer, I just talk to the duck. I’ll narrate what I think Claude might say, or vent about why I even need its help. “Okay Ducky, if Claude ever responds, here’s what we’re gonna do…” On particularly loopy days, I imagine the duck is Claude’s avatar and give it pep talks: “Come on, buddy, you got this – just a little more processing!” It’s goofy, but weirdly it keeps me focused on the problem instead of wandering off entirely. Plus, the duck never ever judges my impatience. Documentation Dive – Here’s a quasi-respectable one: actually reading documentation. Gasp! Shocking, I know. If I’m waiting for Claude to summarize an API or explain a bug, sometimes guilt kicks in and I think, “I could just RTFM myself while I wait.” So I do – I navigate to the official docs of whatever library or tool I’m dealing with, and skim away. This is how I often end up finding obscure footnotes or new features that Claude might not mention. It’s not the worst use of time, honestly. But true to form, the second Claude finally responds with an explanation, I’ll drop the docs like a hot potato. Heaven forbid I continue the manual research when my AI oracle has spoken! Flow Check: Is It Me or You? – There’s a special kind of paranoia that sets in after Claude has been silent for a while. I start wondering if I did something wrong. So I’ll double-check my prompts, re-read the last thing I sent. Did I confuse it? Too much info? Sometimes I’ll even open a separate AI chat (shh, don’t tell Claude) like ChatGPT or Bard to ask, “What would you do with [my question]?” Just to compare. Essentially, I open a second front in the AI war, seeing which model answers first. It’s ridiculous – like hitting the “close door” button on an elevator repeatedly. But it gives me the illusion of control. Worst case, I get a head start from another model; best case, Claude answers and I feel like I set a backup plan in motion. Of course, juggling multiple AIs at once comes with its own mental context-switching cost… but that’s a later realization. In the moment, it feels proactive! I am productive! Slack Surfing – I pop open Slack and scroll aimlessly through channels, reacting with emoji here and there, chiming in on random threads (“Haha yes, totally!” to a joke in #random, or “Good point, we should consider that” in a project discussion I half-followed). The goal is to seem present and engaged at work. This is pure performance because mentally I’m half-absent, my brain still stuck on the problem I asked Claude to solve. I’ll even start drafting status updates: “Working on X, just ironing out Y,” to post later – essentially narrating progress I intend to make once Claude comes back with results. Slack is both a distraction and a stage during these times. I know I’ve truly spiraled when I start scrolling our company’s #announcements channel from last week for lack of anything better to do. The VS Code Peek-and-Poke – Sometimes I’ll stay in my editor and do ultra-menial clean-up. I’ll comb through code files fixing typos in comments, re-alphabetizing import statements, or adjusting indentation to be perfectly consistent. These are things nobody asked me to do and that probably make zero difference, but it’s code-related busywork that feels satisfying. It’s basically tidying my room while I wait for the genius to call me back. On one occasion, I even started renaming variables to more pleasing names (only to later regret it because Claude’s suggestion used the old variable names… sigh). On the bright side, at least my codebase comes out a tiny bit cleaner after each of Claude’s longer daydreams. Existential Tech Crisis – And finally, when the wait is really long and I’m really tired, I descend into a mini existential crisis. It starts innocently: “Hmm, what’s taking Claude so long?” That becomes, “Is Claude… stuck? Did I break it? Do I even know what I’m doing?” Which snowballs into, “Why am I even in this field? Am I about to be outdone by a machine? What’s the meaning of a developer’s work in the age of AI?” Next thing I know, I’m questioning the trajectory of my career, the future of humanity with AI, and whether I should have listened to my college roommate and gone into consulting. This is usually when I realize I’m spiraling – fully spiraling, as the kids say. The funny thing is, the moment Claude finally responds, all those grandiose worries evaporate as I excitedly read the answer. Crisis postponed, back to coding. (I sometimes imagine Claude purposely delaying just to mess with me: “Let’s see if we can induce a minor identity crisis this time, mwahaha!”). The Tab Tornado – I open, oh, about 27 new browser tabs on a whim. Documentation I’ve been meaning to read, that interesting blog someone posted in Slack, random Stack Overflow questions tangentially related to my work. My Chrome looks like it’s cosplaying as my brain: a million things open at once. Naturally, I’ll only actually read 2 or 3 of these tabs (and forget about the rest), but for that brief moment, I feel busy. Pro tip: more tabs = more productivity, right? (No. The answer is no. But in the moment, it sure feels like maybe.) The Tab Tornado busy maybe Micro-Side-Project Mania – In a fit of “I’m totally using this wait time constructively,” I sometimes spin up a tiny side project. This could be as trivial as writing a quick script to automate a personal task, or as over-ambitious as saying “I’ll just start building that app idea I had.” I’ve literally generated a new React app and begun tinkering while waiting for Claude. By the time Claude returns with my answer, I’ve forgotten what I was doing in the main project because I’m knee-deep in a completely unrelated coding escapade. It’s the developer equivalent of stress-baking a cake while procrastinating on writing a report. Micro-Side-Project Mania while waiting for Claude completely unrelated The Snack Expedition – If all else fails, wander to the kitchen. I can justify this because technically I’m optimizing my personal fuel levels, right? I’ll grab a coffee, then think, maybe I need a snack. Next thing I know, I’ve crafted a seven-layer peanut butter and Nutella sandwich (don’t ask) or eaten half a bag of pretzels while staring into space. I’ve had days where my caffeine intake correlated disturbingly with Claude’s response times – the slower the model, the more espresso shots consumed. The outcome: a very jittery developer who has answers from Claude but also hands shaking too much to type properly. The Snack Expedition answers VS Code Theme Makeover – This is a classic “productive” procrastination move: tweaking my development environment. Did I need to download three new Visual Studio Code themes and A/B test which one gives me the best vibe? Of course not. Did I do exactly that while Claude was churning on a particularly long response? Absolutely. There is a special kind of nerdy joy in telling yourself that changing your editor’s color scheme to Nordic Ice or Neon Vamp will somehow make you more productive. It won’t, but it will beautifully consume 10 minutes. Plus, if anyone glances at my screen, they’ll think I’m just super detail-oriented about my tooling (instead of simply desperate for any task to focus on). VS Code Theme Makeover need best vibe Nordic Ice Neon Vamp any Inbox Zero Heroics – Normally, I treat my email inbox like radioactive waste (look but don’t touch unless absolutely necessary). But in moments of AI-induced downtime, suddenly I’m an email-clearing machine. Newsletter I never read? Unsubscribe! Some auto-generated alert? Archive! I even start writing overly prompt replies: “Apologies for the delay, see my answers inline…” – as if the 4 minutes it sat unanswered was a grand inconvenience to the sender. The irony: I’m only ultra-responsive because I’m waiting on Claude. If my colleagues knew my sudden email diligence was a side effect of AI slowness, they’d chuckle. But hey, at least for a moment I feel on top of my correspondence – a veritable Inbox Zero Hero, cape and all. Inbox Zero Heroics 4 minutes because Rubber Duck Debugging (Literally) – This one’s borderline embarrassing, but here we are. I have a rubber duck on my desk (a classic programmer trope) for debugging by explaining problems out loud. Sometimes, while Claude is cooking up an answer, I just talk to the duck. I’ll narrate what I think Claude might say, or vent about why I even need its help. “Okay Ducky, if Claude ever responds, here’s what we’re gonna do…” On particularly loopy days, I imagine the duck is Claude’s avatar and give it pep talks: “Come on, buddy, you got this – just a little more processing!” It’s goofy, but weirdly it keeps me focused on the problem instead of wandering off entirely. Plus, the duck never ever judges my impatience. Rubber Duck Debugging (Literally) think Claude’s avatar Documentation Dive – Here’s a quasi-respectable one: actually reading documentation. Gasp! Shocking, I know. If I’m waiting for Claude to summarize an API or explain a bug, sometimes guilt kicks in and I think, “I could just RTFM myself while I wait.” So I do – I navigate to the official docs of whatever library or tool I’m dealing with, and skim away. This is how I often end up finding obscure footnotes or new features that Claude might not mention. It’s not the worst use of time, honestly. But true to form, the second Claude finally responds with an explanation, I’ll drop the docs like a hot potato. Heaven forbid I continue the manual research when my AI oracle has spoken! Documentation Dive Gasp! manual research Flow Check: Is It Me or You? – There’s a special kind of paranoia that sets in after Claude has been silent for a while. I start wondering if I did something wrong. So I’ll double-check my prompts, re-read the last thing I sent. Did I confuse it? Too much info? Sometimes I’ll even open a separate AI chat (shh, don’t tell Claude) like ChatGPT or Bard to ask, “What would you do with [my question]?” Just to compare. Essentially, I open a second front in the AI war, seeing which model answers first. It’s ridiculous – like hitting the “close door” button on an elevator repeatedly. But it gives me the illusion of control. Worst case, I get a head start from another model; best case, Claude answers and I feel like I set a backup plan in motion. Of course, juggling multiple AIs at once comes with its own mental context-switching cost… but that’s a later realization. In the moment, it feels proactive! I am productive! Flow Check: Is It Me or You? I Slack Surfing – I pop open Slack and scroll aimlessly through channels, reacting with emoji here and there, chiming in on random threads (“Haha yes, totally!” to a joke in #random, or “Good point, we should consider that” in a project discussion I half-followed). The goal is to seem present and engaged at work. This is pure performance because mentally I’m half-absent, my brain still stuck on the problem I asked Claude to solve. I’ll even start drafting status updates: “Working on X, just ironing out Y,” to post later – essentially narrating progress I intend to make once Claude comes back with results. Slack is both a distraction and a stage during these times. I know I’ve truly spiraled when I start scrolling our company’s #announcements channel from last week for lack of anything better to do. Slack Surfing performance intend The VS Code Peek-and-Poke – Sometimes I’ll stay in my editor and do ultra-menial clean-up. I’ll comb through code files fixing typos in comments, re-alphabetizing import statements, or adjusting indentation to be perfectly consistent. These are things nobody asked me to do and that probably make zero difference, but it’s code-related busywork that feels satisfying. It’s basically tidying my room while I wait for the genius to call me back. On one occasion, I even started renaming variables to more pleasing names (only to later regret it because Claude’s suggestion used the old variable names… sigh). On the bright side, at least my codebase comes out a tiny bit cleaner after each of Claude’s longer daydreams. The VS Code Peek-and-Poke nobody old sigh Existential Tech Crisis – And finally, when the wait is really long and I’m really tired, I descend into a mini existential crisis. It starts innocently: “Hmm, what’s taking Claude so long?” That becomes, “Is Claude… stuck? Did I break it? Do I even know what I’m doing?” Which snowballs into, “Why am I even in this field? Am I about to be outdone by a machine? What’s the meaning of a developer’s work in the age of AI?” Next thing I know, I’m questioning the trajectory of my career, the future of humanity with AI, and whether I should have listened to my college roommate and gone into consulting. This is usually when I realize I’m spiraling – fully spiraling, as the kids say. The funny thing is, the moment Claude finally responds, all those grandiose worries evaporate as I excitedly read the answer. Crisis postponed, back to coding. (I sometimes imagine Claude purposely delaying just to mess with me: “Let’s see if we can induce a minor identity crisis this time, mwahaha!”). Existential Tech Crisis I spiraling Whew. So those are my 11 tactics – if you can call them that – to cope with Claude’s thinking breaks. Are any of them ideal? Absolutely not. Effective? Debatable (leaning toward nope for most). Hilarious in hindsight? I sure hope so, otherwise I just admitted to an awful lot of questionable behavior for nothing. But hey, I never claimed to be handling this well – I’m just handling it, period. 11 tactics nope The High Cost of Constant Context-Switching By now it’s pretty clear that my attempts to “stay productive” while waiting are a bit... scattershot. If you’re thinking, “This sounds like it would totally destroy your focus,” congratulations, you win a prize. 🎉 The truth is, constantly context-switching (even in the name of productivity) has a cost. In fact, researchers have found that the worst interruptions for getting work done aren’t the external ones – they’re the internal, self-inflicted ones(blog.stackblitz.com). Reading that hit me hard: I am my own worst enemy whenever I abandon my task to juggle 13 mini-activities during a 60-second pause. constantly context-switching blog.stackblitz.com I It’s the paradox of our times: I fight so hard to preserve my “flow state” from big disruptions, yet I’ll voluntarily shatter it into a million pieces the second I’m left with a moment of idle time. A developer on Reddit perfectly described the dilemma of those short waits: if a build takes ~1 minute, it’s “too long to just stare at the screen, but too short to actually do something” meaningful (reddit.com). That’s exactly the purgatory I find myself in with Claude. One moment of pause is all it takes for me to alt-tab away, and once I’ve broken away, it’s so hard to snap back. Often I’ll end up spending 4-5 minutes on a tangent even though Claude replied after 30 seconds. I effectively create a bigger delay for myself than the AI ever did! reddit.com so bigger This isn’t a new lesson, by the way. It’s basically a remix of the age-old programmer wisdom: don’t break a coder’s focus. Even a tiny interruption, like a “quick” 10:30am meeting, can blow up the whole morning’s productivity – there’s a meme that shows a dev’s schedule: nothing gets done before or after that 10:30 “super quick” meeting because the flow is wreckedcheezburger.com. Exhibit A below: the struggle is real. don’t break a coder’s focus cheezburger.com Exhibit A below: A single “super-quick meeting” in the middle of the morning can derail a developer’s flow for hourscheezburger.com. Waiting for Claude can have a similar effect on my day – a brief pause that spirals into a longer derailment. A single “super-quick meeting” in the middle of the morning can derail a developer’s flow for hourscheezburger.com. Waiting for Claude can have a similar effect on my day – a brief pause that spirals into a longer derailment. cheezburger.com When Claude goes on a 30-second thinking spree, I essentially interrupt myself. I initiate the context-switch; Claude merely provided the temptation. And every context-switch has a re-entry cost – the time to remember where I was, reload the mental state of the code, re-focus my brain. Sometimes I handle it okay. Other times I feel like my head is a shattered snowglobe of thoughts, and I’m desperately trying to get all the flakes to settle back on the original scene. interrupt myself It’s humbling (and a bit amusing, if I detach and observe it academically). I have this super-advanced AI at my fingertips, yet my primitive human brain still struggles with staying present. There’s even research on “fragmented thinking” suggesting that too much task-switching is the silent killer of developer productivityblog.stackblitz.comblog.stackblitz.com. You’d think waiting for an AI would be a passive break, but oh no – I manage to turn it into a fragmentation field. I basically throw my concentration out the window at the slightest provocation. staying present blog.stackblitz.com blog.stackblitz.com Real talk: I probably need to practice just doing nothing for those short spans, or doing something truly refreshing like standing up and stretching, rather than manically trying to fill every second. But that’s a lesson I’m still learning. For now, the struggle (and the juggle) continues. doing nothing Confessions of an Impatient AI Addict At this point, you might think I’m fed up with Claude. That I’m ready to dump my AI companion for a faster model or revert to old-school methods. Let me confess: I’m not. I’m addicted to this thing. For all the grumbling I do during the waits, I still light up like a kid on Christmas when Claude finally delivers a brilliant solution or speeds through a task that would’ve taken me an afternoon. It’s very much a love-hate (or rather love-frustrate) relationship. I vent about the delays, yes, but the second it gives me what I need, all is forgiven and I’m back to singing its praises in the company tech channel. addicted If anything, this whole experience has made me more self-aware… and a bit self-deprecating (as if you couldn’t tell). I recognize my own absurdity in this dance. I mean, who else would open a dozen tabs and philosophize about the meaning of work, all because an AI took 60 seconds instead of 6? It’s almost comically human of me: I found a way to be impatient about a miracle. “Hey, this machine is performing intellectual feats for me that were science fiction a couple years ago – why isn’t it faster!?” Truly, we are difficult creatures to satisfy. self-aware impatient about a miracle I sometimes imagine how this reads to someone in another field. A lawyer or a teacher might say, “Wait, you’re complaining that your robot helper takes a minute to do a complex task? Seriously?” And I’d have to sheepishly reply, “Yeah… I know, I know.” It’s a classic tech worker problem – a 1% problem if there ever was one. I’m literally complaining about waiting for answers that would have been impossible to get so quickly a short time ago. This perspective doesn’t escape me. It’s actually healthy to remind myself of it on occasion, as a reality check. robot helper 1% problem Still, when you’re in the thick of it, trying to keep a flow going, even small delays can feel huge. It’s like being in a high-speed car chase and hitting every red light – sure, you’re still covering ground faster than walking, but ugh, those red lights! The emotional rollercoaster is real. And so the coping mechanisms and mental gymnastics continue, for better or worse. ugh At the end of the day, I do have to hand it to Claude: it’s made me confront my productivity habits in a new light. I’ve learned just how prone I am to “productive” procrastination and how easily I can justify any distraction as “keeping the flow going.” I’ve also learned that maybe – just maybe – I need to chill out a bit. Not every ten-second gap needs to be filled. It’s okay to pause, breathe, maybe even stare out the window (what a concept!) instead of compulsively alt-tabbing. Heck, some developers keep an actual book by their desk for compile waits (reddit.com). Perhaps I could channel a bit of that Zen. (Knowing me, I’d probably choose a book about AI and then end up more riled up, but it’s a thought.) “productive” procrastination reddit.com So here I am: an anonymous tech worker who could plausibly be at Google (I’ll never confirm or deny 😉), pouring my heart out in a rogue confessional about my Claude addiction and the antics I pull to cope. Writing this has been cathartic – and yes, I fully started it during one of Claude’s slower responses. Talk about meta-productivity! Instead of just twiddling my thumbs, I’ve created a whole blog post. Take that, productiveness theater – I turned you into actual production. that Embracing the Wait (Or at Least Making Peace with It) I wish I could end this with a grand solution: a revelation that I’ve mastered patience or a hot tip to eliminate Claude’s wait times completely (Anthropic, if you’re reading, I wouldn’t mind that at all). But the truth is, this is likely the new normal – at least for a while. As we push the boundaries of what these AI models can do, we’re also pushing our own boundaries of how we work alongside them. There will be hiccups, slowdowns, moments where the human is left hanging while the silicon brain grinds away. My saga of waiting is just one colorful example. how we work alongside them What I can offer is solidarity and a bit of hard-earned perspective. If you find yourself spiraling while waiting for an AI (or frankly, waiting for anything in this high-speed digital world), know that you’re not alone – and that it’s okay to step back and smile at the absurdity of it all. We are, in the grand scheme, ridiculously fortunate to even have these problems. I bitch about Claude, but I wouldn’t give it up; I just need to adapt. Maybe I’ll actually close a few tabs, or allow myself to do nothing for a minute (imagine that). Maybe I’ll treat those waits as tiny mindfulness breaks – the universe forcing me to slow down. (Cue eye twitch – I’m not there yet, but maybe someday.) nothing Cue eye twitch there In the meantime, I’ll keep opening tabs and writing snarky commentary in my head to entertain myself. I’ll keep finding the humor in my own impatience. Like a true Casey Newton/Kara Swisher spawn, I’ll call out the nonsense (mostly my own) with a wry smile and a clever punchline. And when Claude finally does finish loading that answer, I’ll happily get back to work – until the next wait, when the cycle repeats. So, what to do while I wait for Claude? The answer, it seems, is anything and everything… and occasionally, absolutely nothing. It’s a work in progress, much like the technology itself. And if all else fails, well, I can always write another blog post about it. After all, you’ve just read one – born in a moment of procrastination, but hopefully ending on a note of genuine insight (wrapped in humor). If that isn’t productiveness alchemy, I don’t know what is. So, what to do while I wait for Claude? Now if you’ll excuse me, Claude just pinged with a response. Time to actually be productive – no theater required. Until the next wait, dear reader… hang in there, and maybe close a couple of those tabs. We got this. 🚀