AI Is Taking Over The Intern's Job

If your job involves thinking, something interesting is happening, and it's important to notice.

Just a few days ago, I observed a junior analyst putting together a market brief for a meeting. In less than twenty minutes, she summarized and extracted key data from twelve different sources, created a clear outline, drafted presentation slides and created speaker notes. Most people would take one whole morning to accomplish that.

What surprised me the most was the absence of any acknowledgment from her peers. No one commented, “That’s incredible.” Everyone seemed to consider that benchmark modern market brief as just average.

The Shift

Surprisingly, most people today still consider using AI to be experimental. Writing a draft, going back and revising a few sentences, and generating a few concepts may feel like innovation for some.

Now, I witness the greatest strategies created easier and faster than anyone I know because of AI. I’ll be in a meeting sharing an analysis of a summarized presentation created by AI. In many cases, the bottleneck is no longer the research or the data but the analysts themselves.

It seems to me the only difference is starting with AI and starting without it. Some people use it to sort of brainstorm while others just get to a point where they can't progress without it.

Over time AI will flatten a lot of the advantages people rely on today. The shift will be slow, but it will be noticeable.

Initially, the impact might only be felt by people with the same backgrounds creating work at a far better level. Then the organization adjusts the baseline to a higher level so the old impressive work becomes a new standard.

To survive and excel in this new normal, this is what we want to do:

1. Restrict the scope of the task in the beginning

Some examples include drafting research summaries, preparing updates for the team, organizing notes prior to meetings, and drafting proposal outlines.

Do the task one time as you normally do and identify where you encounter friction.

Once you do that, you can then improve that workflow with AI. Instead of open-ended requests, you can use structured requests. They can define the type of response, the constraints, and the audience.

Consider the interaction as part of a system that can be repeated, and not just a chat for the “brilliant” answer. The goal of the task is to minimize friction.

If AI can be used to get a strong first draft, the role of the individual is now going to shift from being the primary creator of that content to being an editor.

2. Build a Deliberate A.I. Stack

The better way to use AI is to build a small stack that resembles how your actual workflow goes.

A small example is using ChatGPT or Claude for drafting and structuring the first drafts, using Perplexity for pulling research and summarizing it quickly, and using Notion or Obsidian to store your prompts, outlines, and templates that you can then reuse for all the workflows that you do not want to reinvent every week.

When that framework is in place, simple automations begin to make sense. Using Zapier or Make, for example, can allow meeting notes to be pushed to Notion, have an AI summary generated, and an update draft created, which can be adjusted later.

Again, the focus is not tool collection but the flow. Information arrives, is adjusted, and becomes a draft usable to the user without manually completing each process.

When that system is seamless, there is an increase in output without an increase in the length of the day.

3. Embrace the Possible Beyond Prompts

AI is typically used in a very basic way. Users prompt for a result and if it's good enough, they'll start a new AI session and prompt again.

This is not the most efficient way of using AI, however, as it is much better to structure the work.

Using AI for the initial structure is one efficient way of working. By breaking the work down and using AI to generate as much of an initial draft as possible, users can then adjust the most important parts, such as the argument, details, and tone. It may seem like a minor adjustment but is incredibly impactful to the pace of work.

Unlike trying to build things from scratch, in this case, you are reframing what already exists. Gradually, the reframing process takes less time than most people anticipate.

So... This Is The New Normal?

Drafting with AI, the automation of repetitive thinking activities, and the amplification of output without the extension of work time are becoming part of normal day-to-day activities.

A reasonable approach is to not make sweeping changes, but to start with a limited number of changes. Begin with a single task that you have to perform repeatedly and change how you work. Create a new workflow and use it for a week, and you can improve it at any time.

Currently, the difference between people integrated AI into their workflows and people who don’t is not that big yet. But over time that difference will be rather obvious.

The tasks of the intern have already changed. Right now, the question is very simple. Are you still playing around with AI from time to time, or are you slowly restructuring how you work in relation to it?