The Bug Is the Brief

I built a bot for my wife recently, which sounds more impressive than it was.

Michelle will often ask me some version of, "When are you traveling next week?" Reasonable question. Extremely reasonable question — the kind of question a spouse should be able to answer without opening three calendars, two email threads, and whatever part of his brain stores airport codes. And for years, I could not answer it cleanly.

Her asking didn't annoy me. My not knowing annoyed me. That distinction matters.

At some point, I filed it away. Not dramatically — I didn't open a Notion database called "Domestic Operational Excellence" (though, now that I say it…). I just noticed the friction.

Then, during a workshop where other people were building agentic tools, I built a wife-briefing bot. It reads my calendar, pulls out my travel details, and sends Michelle a short Saturday morning email with the coming week's travel schedule. Not a unicorn, not a moonshot, not a pitch deck. Just a little system to solve a little bug that had been sitting in our marriage for roughly twenty years.

When I tested it, I wasn't even sure it worked — the connection between calendar, email, and LLM can be funky (technical term). Then, about a minute later, I got an all-caps email from Michelle.

Michelle is not an all-caps gal.

It said: THIS IS AMAZING WHY DID IT TAKE YOU 20 YEARS???

Fair question.

Nobody asked me to build it. No one assigned it. No one wrote the spec.

The bug wrote the spec.

And I think that may be one of the most important shifts in the AI era. For years, I treated that annoyance as normal life. Of course I don't know my travel details. Of course Michelle has to ask. Of course I have to dig around. Of course this is mildly irritating forever. Then AI made the solution easy enough that the real constraint became visible. The problem was not that I couldn't build the system. The problem was that I had never treated the annoyance as a brief.

The premium is moving upstream

Most conversations about AI still focus on problem solving. Can AI write the memo? Can AI code the app? Can AI summarize the research? Can AI build the prototype?

Yes. Increasingly, yes.

That doesn't mean problem solving disappears — but it does mean something important changes. For most of our lives, problem solving was the heroic skill. School rewarded us for solving assigned problems. Work rewarded us for executing assigned projects. Clients brought briefs, bosses set priorities, teachers wrote exams, job descriptions told us what role to play. Someone else named the problem. Then we proved ourselves by solving it.

But when AI can help almost anyone solve a well-defined problem, the value moves upstream. The defining skill of the AI era may not be problem solving. It may be problem finding: the ability to notice what deserves attention, turn it into a brief, and commission yourself to act.

That last phrase matters.

Commission yourself.

Agency is not just "working hard" — plenty of people work hard on things someone else assigned. Agency is the ability to say, "This bugs me. This matters. I'm going to do something about it." Before there is a task. Before there is a project plan. Before there is permission. Before there is a manager saying, "Can you take this on?"

Real talk? That is a very different muscle.

Before you keep reading, try something. In the next ten seconds, finish this sentence:
It bugs me that ______.

Not a global problem. Not a strategic priority. Something ordinary. The weekly meeting nobody prepares for. The form customers keep filling out wrong. The email you rewrite every Monday. The thing your spouse asks because you never have the answer ready.

Don't solve it yet. Just name it.

That's a brief. You just didn't know it.

We were trained to wait

A mom recently asked me how she should advise her high school and college-age sons. What should they study? What should they avoid? What should they be thinking about?

I told her there has never been a better time to be young in terms of opportunity.

She pushed back, rightly. There's a real mental health crisis. Lots of young people are anxious, pressured, pessimistic. That's true. And it's also true that opportunity has never been more abundant. Both things can be true.

The differentiator, I told her, is agency. Her sons need to become highly agentic people. High agency, low victimhood. Not because "mindset" magically solves everything (please don't put that on a mug), but because the world increasingly belongs to people who can determine what should be done before someone tells them what to do.

This is obvious with young people, but it isn't only a young-person problem. Students spend years with teachers saying, "Here's the assignment." Then they graduate and start looking for a manager to play the same role: tell me what to do, tell me what matters, tell me what success looks like.

But adults do this too. We wait for the boss to define the project. We wait for the client to name the need. We wait for the meeting to tell us what the meeting was for. (I wish that last one were a joke.)

Somewhere along the way, many of us got very good at asking, "What do you need from me?" and less practiced at asking, "What needs doing?" AI makes that posture expensive. If execution is getting cheaper, waiting for assignments is a terrible strategy.

Most AI advice starts one step too late. It teaches people how to write better prompts. Useful, yes — but prompts are downstream of briefs, and briefs are downstream of noticing. Most people are asking How do I prompt better? The better question is What is worth prompting about? The best question is What have I noticed that deserves a prompt?

The prompt before the prompt is noticing what deserves attention.

The complaint is often the clue

I was having breakfast recently with a couple of computer science students who were talking about how hard it is to find a job.

I said something like, "Who wants a job? Find a customer."

They looked at me the way students sometimes look at you when you have briefly stopped speaking English.

One of them mentioned Minecraft. He was frustrated that Minecraft wasn't developing as quickly as it should — there was some feature, some direction, some possibility he could see as a power user that the company apparently wasn't moving on. My advice was simple: stop applying for a job at Minecraft. Build the thing you wish Minecraft would build. Then take it to them and say, "By the way, it will cost you a quarter million dollars to hire me." (I may have rounded up. That's called teaching.)

But the deeper point wasn't entrepreneurship. It was recognition. He had already done the hard part. He'd noticed a real problem — he had taste, proximity, frustration, context, a point of view about what should exist that didn't.

What he lacked wasn't insight.

It was permission.

That's true for a lot of us. We think we are complaining — often, we are problem finding. A complaint is often an unclaimed brief.

AI cannot be annoyed for you

This is the part I think people miss.

AI can do a shocking amount now. It can draft the email, write the code, build the prototype, generate the options, analyze the data, make the first version embarrassingly fast.

But AI doesn't wake up annoyed.

AI doesn't notice that your spouse keeps asking the same totally fair travel question and you keep failing the same tiny domestic test. AI doesn't get irritated that your team updates the same spreadsheet every Friday. AI doesn't wonder why customers keep asking the same confused question. AI doesn't walk through your day and think, "This reimbursement process is absurd."

You do.

That is not a distraction from innovation. Properly handled, annoyance is often the beginning of innovation. The irony of "agentic AI" is that AI still does not have agency in the human sense. It can pursue a goal. It can't decide what is worth wanting. It can't premeditate. It can't care. It doesn't take initiative. It doesn't say, "Hang on, this thing everyone has normalized is actually ridiculous."

That is your job.

Why we don't do this

There are reasons we don't keep bug lists. Worth naming them, because they're the actual blockers — not lack of bugs.

Sometimes it's embarrassment. If this bug is real, why have I tolerated it for so long? (Twenty years, in my case. Michelle's all-caps email was funny because it was true.)

Sometimes it's helplessness. Fine, I noticed it. Now what?

Sometimes it's social risk. Nobody wants to be the person who "just complains."

And sometimes it's identity. The more senior you get, the more pressure you feel to have answers, not annoyances.

But that's backwards. Noticing is not junior work. Noticing is upstream work.

Keep a bug list

Long before anyone was talking about "agentic AI," Bob McKim — one of the great design teachers at Stanford, and the founder of the program that became the d.school — encouraged students to keep a bug list. Not a list of startup ideas. Not a list of impressive insights. A list of things that bugged them.

I love how unglamorous that is. In the most technologically advanced moment of our lives, one of the most useful practices is still noticing what irritates you enough to write it down. Sixty years later, the most sophisticated designers and innovators in the world still do it — because they figured out something the rest of us are now being forced to learn. Noticing what's broken is upstream of everything. It's the senior move, not the junior one.

The idea is simple: write down what bugs you. Not because complaining is noble (it usually isn't), but because friction is information. The things that annoy you, frustrate you, slow you down, or make you say, "There has to be a better way," are often clues to value hiding in plain sight.

Laura D'Asaro, a former student of mine and serial entrepreneur, puts it beautifully:

"The biggest thing is, I'm hyper-aware of problems. Anytime I find myself feeling annoyed, I write down the problem and think about what I might do to solve it."

That's a very different relationship to annoyance. Most people experience annoyance as evidence that the world is broken. Laura experiences annoyance as inventory.

A bug list isn't a complaint journal — it's a brief-generation system. The translation is simple but it isn't automatic: hearing yourself think, "Why is this so broken?" and replacing it with, "What would I try if this were mine?" That move is the whole game. Not "someone should fix this" — "I wonder what the smallest fix would look like."

Bette Nesmith Graham noticed that electric typewriters made mistakes harder to fix. That little bug, combined with her experience painting signs, eventually became Liquid Paper. Now, not every annoyance becomes Liquid Paper — most won't. My wife-briefing bot is not going public. There is no seed round. Please do not email me asking to invest in SpouseOps.ai.

But that's exactly the point. The bug list isn't only for startups. It's for work, for teams, for families, for classrooms — for anyone who wants to stop waiting for the world to hand them a perfectly formatted assignment. The bug list trains attention. And in the AI era, trained attention is leverage.

Commission yourself

Try this for seven days. Keep a running list of everything that bugs you. Every time you think one of these thoughts, write it down:

• Why is this so hard?

• Why do we keep doing it this way?

• There has to be a better system.

• Someone should fix this.

• I should really know this by now.

Don't judge the bugs yet, don't rank them, don't turn it into a personal indictment (I know I would be tempted). Just collect them.

At the end of the week, pick one. Not the most important one — the smallest one. Then turn it into a brief, using this prompt:

I'm frustrated by [specific bug]. Help me turn this into a brief. Design the smallest useful proof I could test in under eight hours. Give me three scrappy versions, the first step for each, and what evidence would tell me whether I'm onto something.

Then build the proof. Not the platform, not the strategy, not the twelve-month roadmap. The proof. Draft the email, mock the form, build the spreadsheet, record the Loom, create the bot. Test it with one person.

The goal isn't to become an entrepreneur overnight. The goal is to practice commissioning yourself — because that's what this moment requires: Not just better prompting, but better noticing. Not just faster problem solving, but better problem finding. Not just waiting for someone else to say, "Here's what matters," but developing the capacity to say, "This matters enough for me to try."

Michelle's question didn't change.

My posture did.

For twenty years, that little bug sat there, disguised as normal life. Then AI made the solution easy enough that the only real question was whether I would notice it and do something about it.

That is the work now.

Don't wait for the assignment.

The bug is the brief.

Related: Keep A Bug List
Related: Stop Making Peace with Pain
Related: Be the Prompt
Related: Take Your Own Job Before Someone Else Does
Related: Hire Yourself

Join over 36,147 creators & leaders who read Methods of the Masters each week.