- Seven ways to use AI for the parts of the job you dread, every one of them naming a specific task and the thing that gets worse while you avoid it.
- Dread is not a time management problem, so time management tools do not touch it. The cost is the elapsed weeks, not the forty minutes.
- Most dreaded tasks already have a clock on them, written into a policy or a payment term nobody read, and the date is one you can look up today.
- Your calendar and your drafts folder know what you are avoiding better than your task list does, because moving a meeting three times is a fingerprint.
- The recurring one is the one to hand off: a dreaded monthly job can arrive nearly finished as an expense report workflow so the only act left is approval.
Using AI for the parts of the job you dread is different from using it to go faster. Going faster assumes you have started. This is about the thing you have not started: the expense claim with eleven receipts in a folder called "receipts," the invoice you keep meaning to send, the renewal form, the quarterly report, the reply that gets harder to write every day you leave it. You know how to do all of them. That was never the problem.
Here is why this deserves its own edition. The cost of a dreaded task is almost never the task. Priya finishes an illustration job on a Thursday, feels good about it, and does not send the invoice. Five weeks later she is in the group that the 2026 QuickBooks Small Business Late Payments Report counts, where 59% of small businesses have invoices overdue by 30 days or more, up from 47% a year earlier, with an average of $17,700 sitting unpaid. Her invoice took eight minutes to write. The five weeks cost her the rent. That ratio, minutes of work against weeks of delay, is the whole subject.
Most listicles answer this with "use AI to be more productive," which is advice for people who have already begun. Every idea below targets the gap instead, the part between knowing you have to do it and doing it, and none of them ask you to feel differently about the task. Each one ends with something you can start this afternoon.
1. Work Out Which Kind of Dread You Have, Because Four Kinds Need Four Different Fixes
"I hate doing it" is four separate problems wearing one sentence, and the fix for each is different enough that guessing wastes months. Research on task aversiveness, summarized well in this Frontiers in Psychology review of procrastination as emotion regulation rather than poor planning, keeps landing on the same handful of dimensions: boredom, frustration, resentment, and the fear of doing it badly.
Those four point in four directions. Boredom means the job is mechanical and should be handed off, because no amount of discipline makes data entry interesting. Frustration usually means the tool is bad, and the fix is a different tool rather than a better attitude. Resentment means the job is in the wrong hands and the real conversation is about ownership. Fear of doing it badly means you need a template, an example of an accepted one, or a check before you send, and the task itself was never hard. Diagnosing which one you are holding is hard from the inside, and it takes about two minutes with something that has no stake in the answer.
Start today: pick the task you have avoided longest and paste this into a chat window: "I keep avoiding [task]. Interview me with one question at a time, at most six questions, then tell me whether this is boredom, frustration, resentment, or fear of doing it badly, and what each of those four would imply I should change." Answer honestly rather than tidily. The verdict decides which of the next six ideas is yours.
2. Read Your Own Avoidance Fingerprints Instead of Your To-Do List
Your task list contains the things you remembered to write down, which is a biased sample: the stuff you are avoiding hardest often never gets written down at all, because writing it down makes it real. The honest record is elsewhere, in the traces avoidance leaves behind.
Those traces are specific and machine-readable. A calendar event you have moved three times. A draft unsent for two weeks. A folder of statements you downloaded and never opened. A recurring reminder you dismiss rather than complete. A thread where you were the last person to say "I'll get to this." Each one is an avoidance fingerprint, none of them appear on a to-do list, and nobody audits them by hand, which is precisely why the audit is worth doing once. This is not the same job as ranking a pile of known problems, which has its own set of moves. This is finding the problems before they reach a list.
Start today: export last month of your calendar to CSV, or copy your drafts folder subject lines, and ask: "Find every event moved more than once and every draft older than ten days. For each one, tell me what it suggests I am avoiding, in one sentence, and whether it has a deadline." You will recognize two of them immediately and be genuinely surprised by one. The surprising one is the point.
3. Find the Clock the Task Is Already On
Almost every dreaded administrative job has an expiry date written into a document you have never read, and that date is the thing that makes delay expensive rather than merely annoying. People do not know the dates, so the task feels infinitely postponable right up until the moment it is not.
The clocks are real and specific. Business expense reimbursement runs on the accountable plan rules in IRS Publication 463, where substantiating a cost within 60 days of incurring it is the safe harbor for a reasonable period, and many employers write that straight into policy as a hard cutoff, after which a reimbursement can become taxable income or simply get refused. Insurance certificates, professional licenses, domain renewals, warranty claims, and dispute windows all work the same way: a date exists, somewhere, in writing.
Marcus runs a two-van HVAC outfit and had four months of fuel and parts receipts in a shoebox. Reading his own company policy took eleven minutes and turned "I should deal with that" into "nine of these expire on the 14th." He did the nine that night and the rest the following week.
Start today: find the governing document for your dreaded task, which is usually the employee handbook, the contract, the policy page, or the vendor's terms. Paste the relevant section in and ask: "What deadlines does this impose on me, what is the exact date for each one given that the expense or event happened on [date], and what happens if I miss it?" Put the earliest date in your calendar with the consequence written in the title. Reading the policy needs no AI at all. Turning twelve pages of it into three dates is where the model earns its keep.
4. Make the Recurring One Arrive Nearly Finished
Everything above is about a single task you are avoiding. This one is about the job you dread on a schedule, which is a different and much better opportunity, because anything that recurs can be waiting for you instead of waiting on you.
The distinction that matters is between a reminder and a draft. A reminder adds to your pile. It tells you about the dreaded thing and leaves the dreaded thing entirely intact, which is why nudge apps stop working after two weeks. A draft removes the blank page, which is where the dread really lives. If the monthly expense claim turns up on the first with the receipts already matched to the card statement, the categories already guessed, and the two ambiguous lines flagged as questions, the job you dread has become a two-minute approval. Nothing about your discipline changed. The starting position did.
This is the item you do not have to keep doing yourself. The assembly work, pulling receipts out of an inbox, matching them against a statement, filling the form in the shape your finance team wants, is mechanical from end to end, which by idea one means it is boredom and should be handed off. That is a standing expense report automation, and the same shape covers the monthly report, the renewal pack, and the recurring client update. The honest limit: approval stays yours, and it should, because the judgment is the only part that was ever yours.
Start today: name the recurring job you dread most, then write four sentences before you build anything. What triggers it, what information it needs and where that information already lives, what finished looks like, and which decisions must stay with you. Hand over the first three. Keep the fourth.
5. Ask What Changed While You Were Avoiding It
You are not dreading the task. You are dreading your memory of the task, formed on the day you first looked at it, and that memory stopped updating the moment you looked away. Weeks later you are still carrying the original estimate around, and it is usually wrong in one of three ways.
Sometimes the job shrank. Dana, who coordinates permits for a small landscaping firm, spent six weeks dreading a revision request from a county office and finally read the thread back in full: two of the four objections had been withdrawn by email in September and a colleague had already supplied the site plan. Forty minutes of work, not a day. Sometimes it grows instead, which is worth knowing today rather than on the deadline. And sometimes it dies: the client goes quiet, the project is canceled, the deadline moves, and the thing you have carried for a month no longer exists. Getting the current state is a reading job across a thread, a system, and a calendar, worth handing over before you commit an afternoon to the wrong version of a task.
Start today: take the dreaded thing, paste in the full thread or the relevant records, and ask: "Summarize the current state of this as of today. What has been resolved since the first message, what is still open, what is now blocked on me specifically, and has anything made this unnecessary?" Compare that answer to the version in your head. The gap is what the avoidance was costing you in accuracy, on top of what it was costing in time.
6. Write Down the Bar You Are Really Being Graded Against
A surprising share of dread is self-inflicted, and not through laziness. It comes from producing to an imagined standard. You dread the grant report because you are picturing a document that will be read closely and judged, when the funder has a four-item checklist and reads the whole thing in six minutes. The gap between the standard you are imagining and the standard being applied is the dread, and it closes the moment the real one is written down.
Most of the time the real bar exists in writing and you have not looked: a form's own field descriptions, the scoring criteria on a tender, a policy's list of what must be included, the acceptance conditions in a contract. Where it does not exist in writing, the next best thing is the last version that got accepted, which tells you the standard more honestly than any guidance note.
The output you want is three or four bullet points, and then permission to stop, because dread thrives on an undefined finish line. For documents you produce on a cycle, the same bar belongs inside a repeatable document workflow so you are not rediscovering it every quarter.
Start today: paste the form, the brief, or the policy section and ask: "What is the minimum this requires to be accepted, as a checklist? Separate what is mandatory from what is optional, and tell me what a reviewer is checking for in each field." Then produce to that list and stop. Not a draft. Stop.
7. Draft the Sentence That Ends the Obligation
The honest closer: some of the work you dread should not be done by you, and should not be done at all. The weekly report nobody opens. The recurring call that replaced an email. The piece of someone else's job that drifted onto your plate in 2024 and never drifted back. If idea one came back as resentment, this is your item, and every other idea in this list is an elaborate way of avoiding it.
What stops people is not courage, it is wording. Nobody wants to sound like they are dodging work, so the message never gets written and the obligation renews quietly forever. This is a good use of a model: not deciding for you, but producing three versions of the sentence, in your own register, with the reason stated plainly and a workable alternative attached. You pick one and send it, and the job disappears rather than moving down the list. This is not deferring something with a dated note, which is a decision to do it later. This is the message that removes the task from existence. Keep it to work: a client, a vendor, a colleague, an obligation with a business reason attached.
Start today: name one recurring obligation you dread and believe is not yours, then ask: "Write three short versions of a message that proposes ending or reassigning this. One that offers a lighter alternative, one that asks who else could own it, and one that asks directly whether it is still needed. Keep each under eighty words and do not sound apologetic." Send the one that sounds like you. Worst case the answer is no, and you have learned the task is genuinely yours, which is its own kind of relief.
Pick One
Do not run all seven. Pick the task that came to mind in the first paragraph, the one you thought of before you finished reading the word "dread," and do idea three on it today. Finding its real deadline takes ten minutes and changes it from a mood into a date, which is the only conversion that reliably gets these things done.
Sunday we report what changed in AI, Wednesday The Loop builds one workflow end to end, and Monday this column hands you the ideas in between. If your dreaded task turns out to be monthly rather than one-off, it is a workflow rather than a willpower problem, and the agent workflow directory has the shape of it already.
Hand over the job you have rescheduled four times
Describe the recurring thing you keep walking past and where its pieces already live, and BYOBot turns the gathering into an agent you own. It arrives nearly finished and waits for your yes.
Frequently Asked Questions
- The difference is where the work goes. Classic advice tells you to change how you feel about the task. These ideas leave your feelings alone and change the task instead, usually by removing the blank page, finding the deadline you did not know about, or producing the thing in a nearly finished state so the only act left is approval. If the dreaded part is a decision, you still have to make it. AI is useful on the parts around the decision, which is most of the elapsed time.
- Then check your own rules before you paste anything, and assume the answer is no for customer records, payroll data, health information, and anything covered by a client agreement. Three safe moves remain. You can describe the shape of the job without the data in it, which is enough for the diagnosis and the step list. You can work on the policy or the form rather than the contents. And you can run the recurring version as an approval workflow inside tools you already trust, which keeps the data where it already lives instead of moving it into a chat window.
- Start with the one that has a clock on it, which is idea three. A dreaded task with a real expiry date is not competing with the other twenty-nine, it is competing with a deadline you can look up. Once the dated ones are handled, rank what is left by what it costs you rather than by how much you dislike it, which is a separate skill with its own moves.
- Several do, and it is worth saying so plainly. Finding the expiry date in a policy document is a reading job you can do yourself in ten minutes. Writing the sentence that ends an obligation is a thing you already know how to say, you just have not said it. AI earns its place in three spots: pulling the avoidance fingerprints out of calendars and folders you would never audit by hand, turning a recurring dreaded job into something that arrives nearly finished, and producing the ugly first version so there is no blank page to stare at.
