In Parallel.

Typically, if you see me away from my desk, there is probably still work happening on my computer. I might be in a meeting while one AI session is helping me work through a new feature for our website, another is developing social content, and another is looking at search rankings, competitors, or content opportunities. Sometimes one of those things finishes while I am still in the meeting. That is fine. It will be waiting for me when I get back.

The most important thing AI is changing about work is not which tool wins, which jobs disappear, or whether the latest model lives up to whatever was promised at launch. It is what one capable person can reasonably be expected to accomplish.

For most of modern work, individual capacity had fairly obvious limits. You had the hours in your day, the skills you personally possessed, and whatever resources your company could give you. You could get faster and better. You could build a team. But there was still a fairly direct relationship between your own time and the amount of work you could produce.

AI is beginning to break that relationship. I am still responsible for the work. I have to explain what I want, review what comes back, correct it, and decide whether any of it is good enough to use. But I am no longer limited to completing one step with my own hands before the next step can begin.

That changes more than speed. It changes how much work can move at once. Once several streams can move in parallel, the standard eventually changes with them.

When More Becomes Normal

I do not think companies are suddenly going to tell employees, “AI exists now, so you should be twice as productive.” That would be a pretty terrible way to approach this, and most leaders could not yet define a reasonable new expectation anyway. The tools are changing too quickly, the quality of adoption varies too widely, and plenty of supposed efficiency gains disappear the moment someone has to fix careless work.

But formal policy is not what changes the standard first. People do.

Consider two employees with similar experience, intelligence, creativity, and work ethic. One completes every task manually and in sequence. The other has learned how to direct several streams of AI-assisted work at once, checking progress, providing context, redirecting weak output, and bringing the pieces together. That does not automatically make the second person a better employee. It does give that person more capacity.

On any single afternoon, the difference may not look dramatic. Over months and years, it compounds. A project that used to take a week may take a day. Something that took a day may take an hour. More importantly, work you never would have attempted because you lacked the time, the technical skill, or the people may suddenly be realistic.

I am seeing that firsthand in my current role. I am working on a website that I could not have built myself a few years ago because I do not know how to write code at the level required to build it. That is still true. I have not secretly become a software developer.

What I can do is communicate what I want. I can explain how I want the site to feel, recognize what is working and what is not, compare options, and give specific feedback until the result gets closer to what I have in mind. AI can handle parts of the execution that I did not have the ability to handle on my own before.

The limitation of “Taylor does not know how to code” has not disappeared. It just matters less than it used to.

That does not mean developers disappear from the process. I can build much more than I could before and get the site close to what I want, but I still need an experienced developer to review the work, make sure it functions properly, and handle the final implementation. The difference is that I am not asking that person to start with a rough description and build every part of it for me. I can focus on what needs to exist, work through most of the decisions, and bring them something much closer to finished.

Once leaders repeatedly see that kind of work, what looked exceptional begins to look possible. Then it looks repeatable. Eventually it looks normal. The people experimenting at the edge change the standard long before anyone writes a new policy.

That rising standard is not automatically good or bad. It can give someone room to take on work that previously sat out of reach. It can also become a new kind of pressure. If a company sees that one person can produce more and responds only by assigning more, the employee has gained capacity without gaining much agency, and the additional capacity becomes a higher quota.

The difference will come down to what happens to the time and capacity AI creates. Does it give someone room to improve the quality of the work, solve a bigger problem, or spend more time with customers? Or does every saved hour simply become another hour to fill? Companies will have to decide what the additional capacity is for and who should benefit from it.

Three Responses to AI

I see three broad reactions to this shift.

The first is fear. If you have spent years learning how to write, design, analyze, research, code, edit, or do any number of things that AI can now perform at some level, it is reasonable to wonder what that means for your job. What I struggle with is the idea that avoiding the technology somehow protects you from the change. If AI is going to reshape your profession, refusing to understand it does not stop that from happening. It just means other people will be learning how to use it while you are not.

The second reaction is a kind of professional pride. Some people are not particularly afraid of AI. They believe their work is too human, creative, original, or sophisticated for it. They know their craft, they can do the work themselves, and they value the fact that what they produce is human.

There are parts of that mindset I respect. People should take pride in their work. There is a lot of awful AI-generated content. Brand, taste, judgment, emotion, and human experience matter tremendously. But using a new tool does not make your expertise less authentic or less valuable. A great creative person using AI should still be more creative than an average creative person using AI. A strong marketer should recognize a hollow strategy. A great writer should know when a sentence does not work. Expertise is what allows someone to direct these tools well and recognize when the result misses the point.

The third group is the one I would bet on. They are curious enough to find out what is actually possible. They do not have to become AI evangelists, and they do not have to believe every new product will change the world. They simply try things.

Sometimes an experiment works. Sometimes it produces complete garbage. Sometimes they spend an hour trying to make something work and learn that the technology is not ready. But now they know something. They begin to develop an intuition for which tasks AI handles well, which ones require close supervision, how much context a system needs, when a workflow can continue without them, and where they have to intervene. That knowledge is difficult to reproduce overnight because it is not knowledge of one product. It is the accumulated result of hundreds of small experiments.

The Individual Contributor Becomes a Manager of Work

I used to roll my eyes at the phrase “AI agents.” For a while, “agent” seemed to become a catch-all for AI doing almost anything, and it rarely seemed clear what made an agent different from a chatbot, an automation, or any other piece of software. It felt like we were personifying the technology without saying what it could actually do. I still do not love the term. Over time, though, the capabilities became easier to recognize, and the distinction began to make sense.

What I understand now is narrower than the name sometimes suggests. A system may be able to carry a task forward without me watching every step, but it is still working toward an outcome someone else chose. The autonomy is in how the work gets done, not in deciding what deserves to be done.

One of the biggest changes I see coming is a shift in what it means to be an individual contributor. You do not need direct reports to begin thinking more like a manager. At any given time, I may have one session working on our website, another helping with social content, another analyzing search rankings, and another researching competitors. I move between them, check what happened, provide feedback, correct the direction, and send each one back to work while I focus somewhere else.

That is management, even if the work is being done by software. You decide what needs to happen, communicate the desired outcome, provide the right context, evaluate the result, and determine when something is good enough to move forward.

The accountability does not move. If something is wrong, I cannot shrug and say the AI made a mistake. I chose to use it. I directed it, and I decided whether the work was acceptable. What changes is where my attention goes.

Most people begin with AI as a better search engine or writing assistant: write this email, summarize this document, make this presentation. Those uses are helpful, but the larger transition begins when you look at your workload and ask which parts actually require you.

There are parts of a project that require my judgment, my relationship with a franchisee, my understanding of the brand, or my opinion. Other parts simply need to get done. Some are routine, repeatable, or time-consuming without needing my full attention every second. The better I become at separating those kinds of work, the more of my time I can put where it has the most value.

Running several sessions at once creates its own problem. Four sessions can produce four drafts, four analyses, or four new directions, but all of them eventually come back to me. I still have to read them, compare them, decide what matters, and fit the useful pieces into the work. If I start more than I can meaningfully review, I have not expanded my capacity very much. I have created a larger pile.

At some point, individual review is not enough. If a team can suddenly produce several times as much work, it also needs clearer standards for what gets reviewed, who reviews it, and what must be true before something moves forward. Otherwise, the additional capacity simply creates a larger and faster-moving quality-control problem.

AI can remove one constraint and immediately reveal the next. The slow part may no longer be producing a first version. It may be deciding which project deserves attention, bringing separate pieces together, or recognizing that something should not move forward at all. Managing more work only helps if I can still make good decisions about it.

That is not an excuse to avoid the work. Delegation has never meant the manager stops caring about the result. The skill is knowing what to delegate, what context to provide, what standard to set, when to check in, and when to take the work back.

Better Tools Raise the Value of Judgment

AI can produce work. It cannot relieve you of having an opinion.

It can give you five headlines in seconds, but someone still has to know whether any of them are good. It can help develop an image or design, but someone still has to shape the direction, judge whether the result looks credible and fits the brand, and decide whether it should actually be published. It can propose a marketing strategy, but someone needs to understand the customer well enough to know whether the strategy makes sense. It can write an entire article, but someone still has to ask whether there is an original thought anywhere in it.

That is why I think taste becomes more valuable as production gets easier. We are already surrounded by mediocre AI-generated content. That does not prove AI is incapable of producing good work. It proves that producing something is different from producing something worthwhile.

I have no problem using AI-assisted imagery or design in the right situation. We do it. That is different from treating AI as the entire creative process. There are cases where AI can help create or refine an image and the result still feels credible to the customer. There are other cases where it absolutely matters how the image was produced and what it is supposed to represent. If you are a food brand posting obviously fake, fully AI-generated imagery, for example, you risk credibility because people expect what they see to resemble what they will actually be served. The technology may be able to make the image. That does not make it the right creative decision.

The same judgment applies to where AI belongs in a relationship. Franchisee support is one of the clearest examples in my role. Someone invests in a franchise partly because they want experienced people they can call when they have a problem. If one of our owners calls because a social account disconnected, they want to send an email newsletter, or they are working through a marketing issue, I do not want the answer to be “go ask a chatbot.” They should be able to call me.

AI can still improve everything around that conversation. I can record and transcribe the call. If I share my screen and walk through a process, the recording can help me turn that walkthrough into a guide. The guide can become part of a searchable knowledge base. The next person with the same issue can have better resources waiting for them.

The human interaction stays human. The surrounding work becomes more efficient. I think that kind of implementation is much more valuable than asking where we can remove another person from a process.

I do not assume that line will always stay in the same place, either. AI will get better at parts of the work I currently keep for myself. If that happens, the answer is not to defend an old boundary simply because it used to make sense. It is to reconsider where I should stay directly involved.

The durable skill is being willing to redraw that boundary. I need to judge the work as it exists today, not preserve a division that made sense six months ago. Human judgment matters, but that does not mean every task we currently put in that category will remain exclusively human. The point is not to protect a list of tasks. It is to keep making an honest decision about what the work requires now.

Curiosity and Communication Become Career Advantages

Agency is sometimes made to sound more complicated than it is. To me, it means taking ownership. You notice something that could be better and do something about it. You do not always wait for someone to hand you a detailed set of instructions. You ask questions, look for gaps, test ideas, and try to understand whether there is a better way.

AI rewards that mindset because the cost of experimentation is getting lower. Maybe there is a task you have done the same way for several years. You know how to do it, it works, and you could probably keep doing it that way for another several years. The useful question is whether it still needs to be done that way. Could part of it run in the background? Could one step be automated? Could you analyze something you normally do not have time to analyze? Could you attempt something that once would have required another team or another skill set?

You will not know unless you try.

Experiments do not have to produce a finished product to be useful. If I spend an hour trying to automate something and discover that the technology cannot do it well yet, I do not necessarily consider that a wasted hour. I understand the boundary better. I know where the process broke and what context was missing. If a new capability appears three months later, I am starting from experience instead of from zero.

AI is moving too quickly for me to think about it three years at a time. Six months can bring a completely different set of capabilities. Someone who experiments consistently during those six months accumulates dozens or hundreds of small lessons. The person who waits for everything to settle is not standing still relative to them. The gap is growing.

The harder thing to catch up on will not be the interface of one particular product. Products change. The harder thing will be years of intuition about how to approach the work.

Communication sits at the center of that intuition. AI exposes how well, or how poorly, you communicate. If you cannot explain what you want, why you want it, who it is for, what good looks like, and which constraints matter, the work will often miss the mark. If your only feedback is “I do not like this,” neither a person nor a system has much to work with.

We usually describe communication as a soft skill. I am not sure that distinction makes as much sense anymore. When you are directing work through AI, your ability to communicate an outcome directly affects the quality of what comes back. It becomes a technical skill in a very practical sense.

In my experience, the way I communicate also has to change from one model to another. Some perform better with detailed instructions and firm guardrails. Claude Fable, Anthropic’s frontier model, for example, often seems to do better when I share the general idea and leave it more room to work through the approach. In that sense, working with models is a little like managing different employees: the same instructions do not produce the best work from all of them. The only way I have found to learn the difference is to use them, see where the work goes wrong, and adjust.

There is a useful side effect. Learning to give AI better context can expose weaknesses in the way we communicate with people. If I become better at defining an outcome, giving useful feedback, and explaining what is wrong with a piece of work, those same skills make me better at working with a coworker, managing an employee, or supporting a franchise owner. AI may force many of us to become better human communicators.

Small Teams Can Move the Hiring Threshold

Small teams may have more to gain from this shift than anyone. There are only so many people available, so a small team has always had to make tradeoffs. AI moves some of those limits. Work that once required another employee, a freelancer, an agency, or more time than the team had available may sometimes be handled internally.

AI has already changed how I think about headcount on my team. If I were working the way I worked several years ago, I would probably be thinking seriously about adding another marketing coordinator sooner. I still believe we will need more people as the brand grows. More franchisees and more locations will create more projects and more people who need personal support. Eventually, another person will be the right investment.

AI has not removed that need. It has changed where the threshold sits by expanding what our existing team can accomplish before another full-time hire becomes necessary.

Moving the hiring threshold does not necessarily mean there is less work for people. It may delay one kind of hire while increasing the need for expertise elsewhere, because more work now makes it far enough to require review, implementation, or a decision.

It has also changed what I would look for in a hire. Familiarity with AI would matter significantly to me, but I would not expect someone to know every model or name fifty tools. The products change too quickly for that to be a useful standard. I would care whether they had been curious enough to experiment. Have they tried to build something, improve a real process, or make part of their work more efficient? Have they had an experiment fail, and can they explain what they learned?

That tells me more than which subscription they pay for. It shows curiosity, ownership over their development, and a willingness to learn without needing someone to tell them exactly what to learn. I would want those qualities on a team even if AI disappeared tomorrow.

Companies should look for the same qualities in the people they already have. Most organizations probably have a handful of employees who are already testing tools, building workflows, and spending their own time figuring out what works inside their jobs. Before turning AI adoption into a massive top-down initiative, leaders should find those people and give them time to teach.

A general presentation telling fifty people that AI is important has limited value when those people do not know what they are supposed to do with it. Watching a coworker take a real task from inside the company and show exactly how they approached it is different. Even better, let the experimenter sit down with someone and ask what they spend their time doing every day. Then look for a real use case together.

Companies can provide tools, education, examples, and support. Eventually, though, the individual still has to try.

The Real Divide

This is ultimately bigger than ChatGPT, Claude, or whichever tool comes next. The products people use today, and the way we use these products, may not be the same three years from now. The durable skill is being the kind of person who sees a meaningful shift and wants to understand it.

AI is making that trait unusually visible because the technology is moving quickly and touching so many kinds of work at once. The people who adapt are not necessarily the smartest people in the room. They may simply be more willing to start before they have all the answers. They try things just to see what happens. They create bad versions before they create good ones. They spend time learning something that may not work, ask basic questions, and change their minds.

That willingness to experiment gives them an advantage over someone who is highly capable but insists on staying inside the methods that previously made them successful.

I do not think the useful warning is “use AI or you are going to lose your job.” That is too simplistic, and it is not the argument I am making. The bigger risk is becoming complacent while the definition of what one capable person can accomplish continues to expand.

Human creativity, judgment, taste, communication, relationships, and expertise still matter. AI gives capable people a way to apply those qualities across far more work than they could previously touch.

The work can now move in parallel, even when our attention is somewhere else. That makes deciding where to put our attention more important, not less. We have to move between several streams of work, recover the context quickly, understand what changed, and decide what needs us next. That kind of context switching takes practice, and there is a limit to how much of it any of us can do well.

Some of the work may happen without us watching every step. That does not make the result any less ours. Before we publish something, present it, or put it in front of a customer, we have to understand the work, be confident it is ready, and be willing to put our own names and the brands we represent behind it. AI can carry the work forward, but it cannot take responsibility for the result. That responsibility still belongs to each of us.

//