Heroically busy.
You feel productive. You produced nothing. Both are accurate.
It's 4:47pm on a Tuesday.
You've sent 38 emails. Sat through five meetings, two of which could have been a text. You had AI draft a status update, summarize a thread you didn't read, and spin up a slide deck nobody asked for. Your Slack has three green-dot fires. Your calendar tomorrow is a wall of blue.
You are exhausted.
And if a stranger walked in and asked, "What did you actually make today?"… you'd have nothing to hold up. No decision that mattered. No problem you cracked. No thing that exists now that didn't exist this morning.
You weren't lazy. You were busy. Heroically busy.
That gap between how drained you feel and how little you produced is the whole story of this letter.
Everyone's bracing for the wrong AI apocalypse.
Open any newspaper: jobs vanishing, mass displacement, governors signing executive orders to brace for the worst. The whole conversation is about subtraction. The robot shows up. Your role goes away. You're out.
Maybe. Eventually. (I wrote a whole letter on why those timelines are mostly self-serving fairy tales. Letter 81.)
But the boring, much more imminent fear nobody is talking about is this:
AI isn't going to take your job. It's going to flood it.
Not by replacing you. By burying you under an infinite, frictionless, self-generating pile of work that looks like work, feels like work, and creates almost nothing of value.
We have a name for that pile. We call it the Busyness Singularity. (Hat tip to Cal Newport, whose reporting on pseudo-productivity is the spine of today's letter.)
To understand why it's coming, we have to rewind to 1955.
A short history of why your work feels insane
In the mid-1950s, Peter Drucker coined the term knowledge work. His job was to explain to American executives that this new kind of worker was nothing like a factory worker. You couldn't hand a strategist a step-by-step checklist and count the widgets at the end of the day.
Which left a question nobody really solved: if you can't count the widgets, how do you know who's working?
The answer we landed on is what Newport calls pseudo-productivity:
Use visible activity as a proxy for useful effort. The more we can see you doing, the more we'll assume you're being valuable.
It was never a good system. It was a good enough system. In the Mad Men era it meant being at your desk for long hours and looking busy when the boss walked by.
Then we wired the office up, and "good enough" started to rot.
Computers gave you a hundred small admin tasks to look busy with. Email turned busyness into a real-time scoreboard. Mobile erased the line between work and not-work, so every moment became another chance to perform effort.
By 2026, Microsoft's own Work Trend Index, built from real use of their office tools, found the average knowledge worker:
Receives 117 emails a day.
Receives 153 Teams messages a day.
Spends 57% of meetings in ad hoc gatherings called with no calendar invite.
Gets interrupted by a message, email, or meeting roughly once every two minutes during core work hours.
Getting interrupted every two minutes is insane. No wonder everyone’s burned out.
None of this was AI's fault. Pseudo-productivity was broken long before the first chatbot. AI just walked into a building that was already on fire and brought a can of gasoline.
Enter AI: the accelerant
Look at what AI is actually doing in most non-engineering jobs right now.
The greatest hits:
Writing long emails.
Summarizing the long emails other people wrote.
Generating slide decks.
Transcribing meetings, then turning the transcripts into "deliverables."
Producing verbose reports nobody asked to read.
Notice what every single one of those is.
They're all pseudo-productive. They're the stuff you do to look busy. No client is paying for the volume of your internal email.
Now run the thought experiment. You take a workplace that already rewards visible activity over actual value. You hand everyone a machine that performs visible activity for free, instantly, at infinite scale.
What happens?
Work becomes a performative button-mashing contest. Who can generate more slop, faster? Soon you're not writing the emails. Your agent writes them. The recipient isn't reading them. Their agent summarizes them. Then their agent drafts a reply, which your agent summarizes for you, which you skim, which prompts you to have your agent fire back.
A digital blizzard of back-and-forth nothing. The density of shallow work approaches infinity. The whole thing collapses in on itself.
That's the singularity. Pseudo-productivity taken to its absurd mathematical limit.
You don't get fired.
You get drowned.
My point: AI doesn't fix the pseudo-productivity problem. It amplifies it. If your job is mostly performing effort, AI doesn't free you. It traps you in a faster, more exhausting, more pointless version of the same race.
The reframe
This was never really an AI problem. AI didn't break a working system. It exposed a broken one.
Which is, weirdly, good news. The cure isn't "use less AI." The cure is to stop letting visible activity stand in for value, in how you're judged and in how you judge yourself.
Five rules. Install them in order. Together they form what I'll call your Pseudo-Productivity Firewall.
We’ve tested this with my team. The results are exceptional.
We work less. We achieve more.
Rule 1 — Plan the week before the week plans you
Every Monday morning, before you touch your inbox, look at the whole week and ask one question: What would create real, non-ambiguous value for my organization this week? Find those one or two things. Block time for them on your calendar like they were meetings with the CEO. If a lower-priority thing is in that slot, move it or kill it.
Why this works: zoom into any moment and the path of least resistance is always something pseudo-productive. There's always one more email to send.
Decide what matters when you sit down each morning, and you'll drift toward looking busy every single time. The week is the right altitude.
Rule 2 — Keep a Value Ledger
Maintain a living document, like a professor's CV, of the genuinely valuable things you actually did. Make sure these are Outcomes, NOT tasks**.** Here's the project I led. Here's the hard part. Here's where I brought judgment a machine couldn't. Here's the result. Update monthly. Share it. Bring it to your quarterly review: "Here's what I shipped last quarter. What should I aim at next?"
This is the Owner-versus-Chaser move from Letter 87, applied personally. Convert an uncapped, ambiguous variable ("how busy does she look") into a fixed, defensible one ("here's exactly what she shipped").
Example. Two people on the same team. One says, "I've been slammed, like 200 messages a day." The other says, "I cut onboarding time from 14 days to 6, here's the data."
In a layoff conversation, one sentence can hurt you. The other can protect you.
Rule 3 — Run the AI Test
Hold each recurring task up to the light and ask one question:
Could Claude or ChatGPT already do most of this for me?
If yes → that's a flashing red sign to move away from it. Not because you should hand it to AI and feel clever, but because if AI can do it, it's no longer where your value lives. Doing more of it makes you more replaceable.
If no → ("I wouldn't even know how to get an AI to do this except at the edges"). That's the gold. Spend more time there.
If your whole day is "attend meeting, get transcript, generate deck, AI-summarize it, have an agent email it out," you feel incredibly busy. But every step is something the machine does or soon will. You've volunteered to compete with AI on AI's home turf.
Rule 4 — Run an upskill project
Have one valuable, rare skill you're actively building. Ideally smuggle it into real work: "Boss, to do this I'll need to learn X." If you can't, carve out 30 minutes a day on slow, steady progress.
The harder and rarer the skill, the further you escape the pseudo-productivity game. When you can do things most people can't, you stop needing VISIBLE BUSYNESS as a proxy for your worth.
Rule 5 — Write
When everyone else is automating their writing, do the opposite. Slow down. Make every email, report, and message clear, concise, and unmistakably human. Spend more time on the writing you do send, and send less of it.
(This is the hard part. I’m struggling with it too. I’ve gotten lazy with writing, and I’m trying to rebuild that muscle.)
I have nothing against using AI for writing, and I use it a lot. The problem is I realized I’m trading an important workout for my brain for a few minutes of “productivity”. And with poor writing comes poor thinking and poor verbal communication. I see more and more people struggling to articulate themselves, and it’s highly correlated with AI use.
So I’m writing more myself… reformatting later with AI, eventually, but the first draft should come from firing my own neurons.
If you manage people: three add-ons
The singularity hits managers twice. Once for your own work. Once because your team is also drowning, and the drowning compounds into ambiguity that eats your week.
Make workloads transparent. One shared document, who's doing what, with a holding queue for things that need an owner. Stop pretending "we all kind of know" from Slack channels and recap meetings is a system. It isn't.
Run docket-clearing meetings twice a week. Thirty minutes, shared doc. New issues go in the docket instead of an instant message. Two or three times a week, you walk through it: handle, queue, assign, kill. This single ritual saves more context switches than any productivity tool ever invented.
Have everyone keep their own Value Ledger. Review monthly. Stop monitoring visible busyness. Start watching whether each person's value log is growing.
One last thing
The singularity is comfortable. There's a real, narcotic comfort in just sitting there clearing notifications and letting AI make the clearing even easier. It feels like progress.
That comfort is the trap. It’s not sustainable. Every quarter it drains you more. You slowly lose your edge because you’re letting the machine think for you.
AI is only as good as the human operating it.
If humans lose their edge, the net value AI brings to society will be negative.
AI would probably become “smarter,” but the total sum of intelligence (“AI + humans”) will likely decline very sharply, and probably in a very unequal way.
Protect your edge and cognitive fitness.
Have a great weekend.
Stay sharp.
— Charafeddine (CM)
