Pick one workflow. Give it one hour. Not the whole business.

To pick one workflow and give it one hour, choose a single repeatable process that already annoys you, set a timer, map the steps you actually do, and mark the one spot where a decision only you can make keeps stalling it. That is the whole job this week. Not the business. One workflow. One hour.

• A workflow is one repeatable sequence of steps that turns an input into a finished result.
• Shrinking the job to one workflow and one hour removes the part that stalls most projects: deciding where to start.
• The part no tool can take is the judgment call sitting in the middle of your workflow, and naming it is the first real move.
• If the start depends on willpower, it will not happen, so attach it to a fixed time and a fixed trigger.
• You do not need the right tool yet. You need to see the flow before you hand any of it away.
• The first concrete step is not solving the workflow. It is mapping one workflow for sixty minutes.

What does it mean to pick one workflow?

To pick one workflow means choosing a single repeatable process in your business, like onboarding a new client or sending an invoice, and agreeing to look at only that one for a set block of time instead of trying to fix how everything runs at once. A workflow is one chain of steps that moves something from start to done. Picking one is the smallest honest unit of work you can commit to. It is not the business. It is one line in the business.

Here is the thing I hear most, and it is almost word for word every time. Someone tells me the tool is fine. The software is fine. The automation half works. Then they go quiet and say some version of this: I mean, that’s when it’s the hardest for me, is when it’s up to me. Right? That moment is the whole post.

The hour is not hard because it is big. It is hard because it is up to you. No tool starts it. No tool decides which workflow matters first. That decision is yours, and that is exactly why it keeps not happening.

The part of a project that stalls is rarely the tool. It is the one decision only a person can make, sitting untouched in the middle of the workflow.

Why one hour and not the whole business?

One hour works because the whole business is not a task, it is a category, and you cannot start a category. You can only start a task. Shrinking the job to one workflow for one hour turns a vague dread into something small enough that you will actually open it before the week gets loud. When the scope stays huge, the start stays impossible. The job defends itself by being too big to begin.

Watch what happens when scope creeps. You sit down to fix invoicing. Twenty minutes in you are rethinking your whole client intake. Ten minutes after that you are shopping for a new tool. The hour ends and nothing is done, because you traded one workflow for the entire business. Less Mess. More Momentum. means you refuse that trade on purpose.

Harvard Business Review has written about what they call the last mile problem, where the real obstacle to an AI project is not the model or the data but the human and organizational work where capability has to meet reality. You can read their take on the last mile problem slowing AI transformation. The last mile is the part that is up to you. One hour is how you walk a few feet of it.

How do you actually spend the hour?

You spend the hour mapping one workflow from first step to finished result, writing down what you really do rather than what you think you should do, and marking the single point where a decision only you can make causes the stall. You are not fixing it yet. You are seeing it clearly for the first time. That is the entire assignment. Sixty minutes. One process. No tool shopping allowed.

Here is the sequence I give people:

1. Name the one workflow. Say it out loud. “Client onboarding.” “Monthly invoicing.” One thing.
2. Set a timer for sixty minutes. When it rings, you stop, even mid-thought.
3. List every step you actually take, in order, including the small annoying ones you usually skip describing.
4. Circle the step where you always pause. The one that waits for you to decide something.
5. Write one sentence about what that decision needs. Not how to automate it. Just what it requires from you.

That circled step is the point. Everything around it can probably be handed to a tool later. The circled step is the part no tool can take, because it needs your judgment, your taste, or your context. That is The Clarity Code in practice. You find the one human decision and stop pretending software will make it for you.

You cannot automate a decision you have never written down. The first hour is for writing it down, not handing it away.

What counts as the part no tool can take?

The part no tool can take is the judgment call inside your workflow. Deciding which lead is worth a real reply. Deciding whether a draft is good enough to send. Deciding what the client actually meant. A tool can draft, sort, and send. It cannot own the call. One useful breakdown of the places where a person has to own the decision inside an AI workflow comes from The Meta Learner’s guide to human judgment in AI workflows, which names defining the problem and deciding which evidence counts as two of them.

Tools versus the human decision

Here is the split, plainly.

Automation platforms like Make.com and assistants like Claude or ChatGPT are good at the left column. The right column is your hour. That is where the project lives or dies.

What happens to a project when that part does not happen?

When the human decision in a workflow never gets made, the project stalls in a half-finished state where the tools are set up but nothing runs, because every automation eventually reaches the point that needs a person, and a person who was never asked to decide cannot start. The project does not fail loudly. It just sits. Half built. Quietly waiting on you.

I have watched this exact thing. The build is done. The connections work. And then it stops, because the one decision in the middle was never named, so no one owns it. The pattern I see in a large share of stalled projects is not a broken tool. It is a decision no one wrote down and no one claimed. HBR has also argued that companies lose AI returns when they automate individual tasks instead of redesigning the workflow itself, which is why looking at the whole flow for one hour beats bolting a tool onto a step.

How do you make sure the hour actually happens?

You make the hour happen by attaching it to a fixed trigger instead of a mood, using a simple if-then plan like “if it is Tuesday at nine, then I open the workflow and start the timer,” so the start no longer depends on feeling ready or finding willpower. This is a known method. The National Cancer Institute documents implementation intentions, which are if-then plans that link a specific situational cue to a specific action. Set the cue. The action follows.

Do not plan the whole week. Plan the one trigger. If Tuesday nine, then workflow. That is it.

Implementation intentions were introduced by psychologist Peter Gollwitzer in 1999, and the format is simply “if situation X happens, then I will do Y.” The power is not motivation. It is that the decision of when and where is made in advance, so the moment arrives already decided.

Almost every time I open a client’s stalled project, I find the same thing. The tool is fine. The wiring is fine. What is missing is one circled decision that nobody named, so it sits in the middle of the workflow waiting for a person who does not know they were supposed to show up. The hour is hard for the same reason it matters. It is up to you, and that is exactly the part no tool can take.

FAQs

How do I pick which workflow to start with?

Pick the one that annoys you most and runs most often. Frequency plus friction means you will feel the payoff fast, which keeps you going. Do not pick the biggest or the most important. Pick the one you keep sighing about. The sigh is the signal. One workflow, chosen, is worth more than ten you are still weighing.

What do I do if I cannot finish the workflow in one hour?

You stop anyway. Finishing the fix was never the goal for this hour. Mapping the workflow and circling the one human decision is the goal, and that fits in sixty minutes. When the timer rings, you have what you came for: clarity on the part that is up to you. The fix is a different hour, on a different day.

Why does it feel so hard to start when it is up to me?

Because there is no external trigger forcing the start, so it waits on willpower, and willpower is unreliable. The fix is to borrow a trigger. Attach the hour to a fixed time with an if-then plan. When the cue is external and specific, the start stops depending on how you feel that morning.

Do I need to choose a tool before I start the hour?

No. Choosing a tool first is how scope creeps and the hour disappears. Map the workflow and find the human decision before you touch any software. The tool question is easier and often smaller once you can see the whole flow. See the flow, then shop. Not the other way around.

What is the single most common reason a workflow project stalls?

One decision that needs a person was never named, so nobody owns it. The tools run until they hit that point, then everything waits. Naming the decision and assigning who makes it is usually the move that gets a stalled project moving again. It is rarely the software.

Can I expand to a second workflow after the hour?

Not this week. The entire point is one workflow and one hour, held on purpose. Expanding scope is the habit that kept you stuck. Do one. Let it be done. Next week you can pick another, with the same rule and the same single hour.

Your next step is small and specific. Pick your one workflow right now. Not three. One. Then set the trigger: a day and a time this week when you will open it and start a sixty-minute timer. Write the if-then sentence down where you will see it. If it is that day at that time, then you open the workflow.

Save this and pick your one. Do it before the week gets loud. The first concrete move is not fixing the workflow. It is choosing the one workflow and setting the trigger. That is the whole job today.