The AI work I’m most interested in

I use AI every day, but the use cases I get most excited about are usually pretty unglamorous.

I’m interested in the report someone rebuilds every month. The research that takes an hour before the actual work can even start. The customer examples everyone knows exist but can never find when they need them. The meeting that contains six good content ideas that will otherwise disappear as soon as everyone moves on to the next meeting.

There is a lot of work around the finished marketing deliverable, and that is where I keep finding some of the best opportunities.

Take something as simple as writing an article. The writing itself is only part of the process. Someone has to figure out the topic, find the right source material, understand what the subject-matter expert thinks, review what has already been published, research supporting information, draft it, fact-check it, get feedback, revise it, format it, and eventually distribute it.

Campaigns, reports, sales materials, webinars, and website projects all have their own version of that.

When I’m looking at where AI could help, I’m usually looking at the whole process:

• Where are people searching for the same information over and over?
• What gets rebuilt every month?
• Which parts of the work require real judgment?
• Which parts are mostly organizing information, formatting something, preparing a first pass, or pulling context together?

That distinction matters.

A good workflow is more useful than a clever prompt

There are plenty of great prompts out there. I use prompts constantly.

But if a task happens again and again, I’m much more interested in building a repeatable way to do it well.

For work that happens regularly, I usually want a few things in place:

• The right source information is available.
• There is a clear sense of what a useful output looks like.
• The team knows what information the tool can and cannot access.
• Someone knows what needs to be reviewed before the work gets used.

That is where AI starts becoming part of how the team works instead of another tool someone opens when they remember it exists.

Reporting

Reporting is a good example. I would rather have a repeatable process where the numbers, definitions, historical context, and business priorities are already organized than ask a chatbot to “analyze my marketing report” every month. AI can help identify changes worth looking at, organize an initial summary, or prepare questions for review. Someone still has to decide what the numbers actually mean.

Research

Research is similar. AI can speed up competitor reviews, organize market information, compare messaging, summarize interviews, and help with the early stages of analysis. That can save a lot of time, but I still want to know where the information came from and whether I trust it before it influences a recommendation.

Content workflows

Content workflows are another obvious one. A good interview, webinar, or subject-matter expert conversation can contain weeks of useful material. Having a consistent way to pull out ideas, organize them, draft different formats, and preserve the person’s actual point of view makes that source material a lot more useful.

I’m equally interested in where AI should stay out

There are tasks where the information is too sensitive for a particular tool. There are outputs where the stakes are high enough that saving twenty minutes is not the priority. Sometimes the process is so unclear that automating it just helps you do the wrong thing faster.

And sometimes AI simply doesn’t make the work better.

I think we should be comfortable saying that.

The best AI use cases I’ve found are usually the ones that quietly remove friction. A recurring task takes twenty minutes instead of ninety. Someone can actually find the information they need. Reporting gets easier to prepare. Good ideas stop disappearing. A process becomes easier for someone else to repeat.

That is less impressive in a demo. It is a lot more useful on a Tuesday afternoon.