In the AI Era, the Hardest Part of Working Alone Isn't Development—It's Continuously Doing the Right Things

Over the past two years, I've increasingly felt that AI has indeed expanded what one person can create.

Previously, many tasks naturally seemed like they required a team. You'd need to write the frontend, write the backend, set up APIs, handle deployment, write documentation, configure workflows, and also manage content, growth, customers, feedback, and iterations. Just thinking about it made it feel like a lot of work with a long chain.

But now it's different.

Many things that used to take much longer can now be done in a first pass with AI's help. Code can get a draft, articles can get a structure, requirements can be broken down, solutions can list several directions, and many repetitive tasks can be significantly compressed.

From the perspective of "can it be built," the barrier is indeed lowering.

But my recent and stronger feeling is:

The hardest part of working alone is not development at all.

The real challenge is:

Continuously doing the right things.

This sounds a bit vague, but my sense of it is becoming more concrete.

Because development difficulties are usually visible. If you don't know, you don't know; if you're stuck, you're stuck. Technical debt, bugs, deployment issues—at least the problems are right there.

But "doing the right things" is different.

The most troublesome part is that often you're not obviously doing wrong. You might even feel like you're always busy, always making progress, always solving problems.

It's only later, looking back, that you realize you spent a lot of time on things that weren't the most worthwhile at the time.

AI Makes "Building" Easier, and Makes "Building Right" More Important

I increasingly think the biggest impact of AI isn't just improving efficiency, but pushing a question to the forefront:

Since many things are easier to build, what exactly should you build?

Previously, many people were limited by their development skills. They had ideas but couldn't build them; they knew what to do but lacked time and energy; one person could hardly handle the entire chain.

So back then, "being able to build" itself was a threshold.

But now that threshold is lowering.

So the questions that come before become more important:

• Is this direction worth pursuing?
• Should this feature be built now?
• Is this the current top priority?
• Should I focus on the product or on delivery first?
• Should I improve the system or validate first?
• Should I expand or consolidate?

AI can't truly make these decisions for you. It can assist, but it won't bear the consequences of your choices.

In other words, the more AI helps you "do," the less you can avoid "choosing."

The Easiest Trap for Solo Workers Isn't Not Knowing How, but Wanting to Do Everything

This is something I feel strongly about now.

When working alone, especially with AI, it's easy to enter a state where:

You can do this, you can try that, you can patch the frontend, you can set up the backend, you can build the admin panel, you can write content, you can automate, you can optimize workflows.

On the surface, this seems like an enhancement of capability. But from the results, it also brings a new problem:

You'll become more easily distracted.

Because when many things are "doable," people naturally overestimate their bandwidth. They think, "I'll do this on the side, patch that on the side, this direction is worth trying, that feature isn't too big."

In the end, it's easy to become:

• Making a little progress on everything
• But not making concentrated progress on what truly matters
• Being busy every day
• But struggling to articulate what key result you advanced this week

I increasingly feel that the hardest part of working alone isn't having few resources, but easily spreading limited resources across too many things that are "not wrong, but not critical either."

The Real Challenge: Direction, Pace, Trade-offs, and Persistence

If I were to break down "the hardest part of working alone," it would be these four things.

1. Direction

Direction isn't "what you want to do," but "what you should do most right now."

These are two different things.

Often, it's not that you lack ideas, but that you have too many. It's not that you have nothing to do, but that there's too much you could do.

At such times, the hardest part isn't pushing forward, but converging.

You need to constantly judge:

• Is this direction a fake need?
• Is this need just something I'm personally interested in?
• Will doing this now bring key changes?
• Is this action advancing results or creating a sense of busyness?

When working alone, if the direction goes wrong, the cost of correction is high. Because no one else helps you correct course in time; often you can only see it yourself in hindsight.

2. Pace

Many things aren't impossible to do, but shouldn't be done now.

This judgment is hard.

For example, does a feature have value? Possibly. Should a system be completed? Eventually. Should a process be standardized? Probably. Is a content direction worth exploring? Maybe.

The issue isn't whether these things are right, but whether they should be done now.

Pace determines whether you complete things one by one or pull everything in early, causing them to conflict.

I increasingly value a "sense of stage."

Is this stage for validation or delivery? For expanding features or tightening boundaries? For accumulating content or solving cash flow first? For long-term building or addressing immediate bottlenecks?

Doing many things too late is a mistake, but doing them too early can also be a mistake.

3. Trade-offs

When working alone, it's easy to mistakenly think, "Doing a bit more can't hurt."

But I've come to realize that's not true.

Often, what really sets people apart isn't how much you do, but whether you dare not to do things that aren't necessary right now.

Trade-offs are painful because they mean giving up. Giving up directions that seem reasonable, giving up attempts you're interested in, giving up actions like "I'll definitely need this later, so doing a bit now is fine."

But without trade-offs, things pile up, attention fragments, and you end up feeling like you're making progress every day, but the core results never truly materialize.

I increasingly believe in this saying:

The prerequisite for doing the right thing isn't just effort, but daring not to do.

4. Persistence

The hardest part of working alone isn't a burst of effort on a given day, but whether you can sustain it long-term.

Many things are easy to start.

The hard part is:

• Are you still doing it in the third week?
• Have you drifted by the sixth week?
• Two months later, do you still remember why you started?
• When real problems arise, does the whole rhythm fall apart?

Persistence isn't about pumping yourself up; it's more like the ability to constantly correct and return to the main line.

My understanding of "persistence" is becoming more realistic:

It's not about charging ahead non-stop, but even if there are fluctuations, you can keep pulling yourself back to the right direction.

AI Can Improve Efficiency, but It Can't Decide Your Priorities

I'm increasingly certain about this.

AI can indeed save me a lot of repetitive work. In writing, development, organizing, breaking down, and drafting, it can already help me cut the first leg of the journey.

But some questions it can't decide for me.

For example:

• What should I focus on most this week?
• Should I continue this feature?
• Is this the most important thing right now?
• Am I validating or self-soothing?
• Is this investment worth it?

In the end, I have to bear these myself.

So I increasingly don't see AI as "someone who completes work for me," but as a tool that constantly pushes me to make clearer priority judgments.

It can help me go faster. But if the direction is wrong, it can also make me go astray faster.

From this perspective, the stronger AI becomes, the more important priority judgment becomes.

My Reminder for Solo Workers Is Getting Simpler

I increasingly feel that the biggest fear for solo workers isn't slowness, but continuously spending limited time on things that shouldn't be done first.

Is development hard? Yes. Is delivery hard? Yes. Content, growth, product, sales—none of these are easy.

But these difficulties are often "visible difficulties."

The more hidden difficulty is:

You're doing things every day, but are you doing what's most important right now?

This question often takes time to answer. And that's why it's harder.

So now I often remind myself:

• Is this the most important thing to do now?
• Am I advancing results or creating a sense of progress?
• Am I solving problems with efficiency, or using efficiency to mask direction problems?

Because for me, in the AI era, the hardest part of working alone is no longer development.

Development is increasingly becoming a "highly assistable" part. But direction, pace, trade-offs, and persistence still have to be handled by yourself.

So now I increasingly ask myself, not:

"Can this be built?"

But:

"Is this worth doing right now?"

Often, the latter question truly determines the outcome.