One workday, compressed: work faster with AI on every task

By the end of one workday, this is what left my hands: 31 emails answered, a client proposal sent, a prospect call prepped, two content drafts ready, invoices out, and the client strategy plan finished for once instead of leaking into my evening. That's my full list, plus the work that never fits. Last Tuesday I wrote down every task in a normal working day, seven in total. On Thursday I ran the same list again, this time letting AI take the first attempt at everything it could possibly touch. If you want to work faster with AI, this is the honest version of what such a day produces: the routine half of my work all but vanished, one task produced a document that could have cost me a client, and two barely changed at all.

Can you really work faster with AI?

Yes, and the proof is in what the day produces, not in racing a clock: my Thursday ended with the entire normal list done plus the deep work that usually gets squeezed out. But the gain landed wildly unevenly. The shallow half of my day (email, planning, meeting prep, first drafts) was finished before lunch with AI doing most of the typing. The deep half (a client proposal, strategy work, admin) still needed me for everything that mattered. Research from Stanford Graduate School of Business points the same direction: AI can make common jobs more productive without sacrificing quality, but the size of the boost depends on the task. My Thursday agrees.

What you need for a compression day

You need three things: yesterday's task list written out in full, one AI assistant you already pay for, and a hard rule that AI gets the first attempt at every task on that list. I used Claude for drafting and research, plus my regular calendar. Total extra cost: $0 on top of the subscription I already had. Spend an evening beforehand writing down what a normal day of yours actually contains, task by task. Skip that step and you'll have nothing to compare against, which turns the experiment into a vibe.

The shallow half: where AI did almost all the work

Email, day planning, meeting prep, and content drafts were all done before lunch, and AI carried the bulk of every one of them. The inbox was the biggest single win: 31 emails, normally a chore I pick at all morning, cleared in one sitting. I pasted each thread in, asked for a reply in my tone, and edited maybe one sentence in three. Planning the day was a solved problem, because I already had the prompt from the week I let AI plan my entire workday. Meeting prep for a prospect call went from my usual tab-hopping spiral to one AI research summary, plus me checking its two most important claims by hand. A LinkedIn post and a newsletter section, normally an exercise in staring, came out of AI first drafts I only had to sharpen.

Nobody pays you for a tidy inbox.

As a freelancer I'd never seen it laid out this plainly: nearly half of what fills my day is this kind of shallow work. Watching all of it get done before lunch stung a little.

Where AI nearly cost me a client: the proposal that invented my rate

The client proposal was the task I expected AI to win, and instead it produced the most dangerous document of my day. The first draft looked ready to send: clean structure, sensible scope, a price. The problem is that I never gave it a price. It had invented a day rate 30% below mine and promised a delivery date I never mentioned. Send that, and the awkward conversation two weeks later starts with my own signature at the bottom. The fix: I fed it my actual rate card plus two old proposals, and told it to leave every number blank unless it came from me. The second draft held up, and the proposal went out the same morning, which for a task that usually camps on my to-do list for days is its own small miracle. Fast, but only after almost being expensive.

The deep half: what AI couldn't touch

Client strategy and invoicing stayed mostly mine, because AI can't do the part of them that matters. The strategy block is the clearest example: the thinking is the work, and AI only helped with the writing-up at the end. Every time I tried to hand over the thinking itself, I got plans that sounded right and said nothing, so I stopped trying and just thought. The plan still got finished by late afternoon, which on a normal day it doesn't.

My accounting software, meanwhile, remains immune to artificial intelligence. AI wrote the polite email wrapping around my invoices; the clicking stayed entirely mine.

What one workday actually produced

By the end of the day, all seven tasks were done, plus the strategy plan that usually spills into my evening. Here's the whole day, task by task:

Bottom line: the list that used to be my whole job became the morning's work, and the afternoon went to billable strategy thinking instead of my inbox. An afternoon of client work at my rate is over $400, and since my costs didn't move, that's just profit. The catch is that you can't run every day at this intensity, because reviewing this much AI output is real work in itself. What you can do daily is the shallow half, and the smarter move there is to stop doing it by hand every morning and automate it once, which is exactly what I did the week after when I automated 5 tasks in 5 days.

When compressing your workday doesn't work

Skip this experiment on days full of meetings, judgment calls, or work you're doing for the first time. Meetings don't compress because other people own the clock. Judgment calls don't compress because reviewing an AI's judgment is just forming your own with extra steps. And new work doesn't compress because you can't check output for a task you've never done yourself, which is exactly how invented day rates end up in proposals. AI works best on the work you already know cold, and that happens to be the work you're most bored by. Convenient, if you think about it.

Would I compress a workday again?

Not as a daily setup, no. Reviewing that much AI output back to back felt like grading a stack of exam papers, and by the strategy block my judgment was noticeably mushier. What stays is the split: my shallow work now runs through AI every morning, and my deep work gets the rest of the day instead of the scraps. Squeezing a whole day's work into as little time as possible was the wrong goal. Ending the day with the full list done and the real work moved forward turned out to be the actual prize.