AI didn't free me up. It made me busier.

The prediction was right. The direction was wrong.

For the past six months, I've been using AI agents heavily — for research, writing, planning, operations, and software development. And I can confirm that it has changed my work. Just not in the way the headlines suggested.

I'm busier than I was before. From my experience working with these tools, any dream that adopting AI creates more leisure time seems unlikely. Here's what I've found so far.

Learning a new skill takes time and repetition

Most obviously, learning to adopt new agent-based workflows takes time. While the tools are improving rapidly and the cost of experimentation is incredibly low, it still takes significant effort to figure out how these tools should be used practically for each specific use case.

Unsurprisingly the result of this has been an overwhelming surge in getting-started guides, developer advocacy content, and even full-blown consulting firms — all created with the specific objective of driving AI adoption in production use cases.

This wave of content is incredibly important because the pace at which the tools, workflows, and best practices are changing requires constant study. The good news here is that this should slow down once well established patterns are cemented and the pace of innovation slows down - if it slows down.

Agents need management, and management is hard

The productivity promise of AI is real, but it comes with a condition that nobody puts in the marketing materials: you have to manage it.

Agents can be autonomous, but autonomy is only good in theory. Just like you would probably not have a new hire on your team run fully autonomously, agents need context, direction, and correction.

And unlike most new hires, agents are prone to confidence in their mistakes. They can interpret instructions both literally and liberally at the precise wrong moments.

Therefore it is critically important to get an agent from "working on something" to "working on the right thing the right way." This requires alignment, and continously managing alignment is hard.

Many great business leaders will immediately recognize this problem. It is a problem the industry has been writing about for eons - organizational strategy, culture, and operations.

Business leaders have long established corporate mission, values, purpose and organizational systems to create a culture in their organizations. This drives alignment at the highest level since the C-Suite will never have the time to manage each and every team member.

When your organization's standard operating procedures hit an edge case, how should that employee intuit the correct series of actions? Agents need this same sort of guidance and that manifests as instruction sets, memory files, and structured project hierarchies.

These instruction sets can and should operate at the highest level (e.g. strategic direction) or lowest level (e.g. standard operating procedures). And just like managing these for an organization today, the work is hard.

Not only is it hard, none of that work shows up in a productivity metric. It's overhead. Critical overhead — but overhead.

People managers understand the leverage you get from a strong and aligned team is real, but it doesn't come for free. You get out what you put into direction, structure, and learning or iteration.

This does not even start to touch on how we evaluate performance and redirect an agent when the systems are insufficient to drive performance. A whole separate blog to come on my experience with that.

Everything feels in-scope now

Here's the other thing that happens when you have AI capacity at your fingertips: your scope expands.

Problems that were previously too expensive or too far outside your skillset become addressable. So you address them. That new machine learning model you wanted to build? Buildable. That competitive brief you never had time to write? Writable.

This is genuinely good. But it means your to-do list grows faster than AI can clear it. You're not replacing work — you're surfacing more of it. AI lowers the threshold for what feels worth attempting, and so you attempt more.

The limiting factor for your to-do list starts moving from your time to your creativity. The result is a fuller plate, not an emptier one.

So should you worry?

Why worry when you can take action? Getting hands on here is the easy part and no one knows your workflows and expertise better than you. If anyone is going to disrupt the way you do things today, the best person in a position to do so is you.

But this doesn't mean just downloading ChatGPT and using generative AI everyday. It means embedding these tools into your workflows. It means learning how to drive alignment for agents. It means learning how to prioritize the important work regardless of whether that work was in-scope previously or not.

The people and teams who figure out how to direct this capacity — how to manage agents, maintain alignment, and translate AI leverage into real outcomes — will have more work to do tomorrow than they had today. That's job security. That's increased productivity. That's new work that wasn't possible yesterday.

That's not a comfortable framing either. It means the work is on you. The work is on us.

We're all busier. But it's an opportunity. Make it count.