How the sausage gets made
Recently, a nerd processor reader asked me why I don't use AI to make nerd processor, since I write a lot about using AI at work.
I replied that of course I use AI to make nerd processor! Just not in the way you might think.
Where I don't use AI
When you read "of course I use AI to make nerd processor," your brain probably went first to the obvious scenario: using AI to write copy. But I don't use AI to write.
I write my own copy for the same reason I always have: Writing is how I work out what I think. If you see public writing with my name on it, on nerd processor or social media or mainstream media or anywhere else, I wrote it myself. To the possible chagrin of every comms person who has ever supported me, I don't use ghostwriters and I don't use AI. Without writing something myself, I wouldn't even know what I wanted to say.
The same is true for my charts and comics. I design and draw them myself.
Where AI makes nerd processor better
I don't use AI to write or draw, but I've used AI elsewhere in making nerd processor. Some use cases have worked better than others.
#1: Understanding my audience (Grade: A)
Kit, my newsletter provider, gives me nice audience engagement stats. For everything I publish, I can see my open rates, click-through rates and specific click-throughs, how many subscribers I win and lose, and more.
Every month or so, I dump all the latest engagement data into my favorite AI along with other metadata: subject lines, newsletter lengths, topic segments, and more. I ask AI to find the patterns that drive success so I can continue to publish things that my audience will enjoy.
For instance:
AI concludes that my engagement stats are strong. I have a 60%+ open rate, 6%+ click rate, unsubscribe rate < 0.2%, so I knew that without AI. (Thanks, y'all.)
AI draws conclusions about my true audience. According to AI, my most engaged audience is "tech execs and senior ICs; AI-curious operators; HR leaders; founders; very online professionals who love data; and people who like learning from someone smart, funny, and sharp-elbowed."
AI concludes that the nerd processor audience skews:
- highly educated
- managerial or aspiration-to-managerial
- deeply interested in work culture
- curious about AI, but not necessarily practitioners
- "fans of your voice as much as your content" (thanks for the pep talk, AI sycophant)
These are things I suspected but didn't know, so I find this insight valuable.
#2: Understanding popular topics (Grade: A-)
AI also gives me insights about the topics that perform best. No surprise, new data analyses, real applications of AI, and discussions of workplace conflict are at the top of the list. And whatever the topic, AI pointed out that my spicy takes perform better than my gentler ones.
AI suggests that I'm at my best when I'm a provocateur. I already suspected this, but nice to have it validated.
#3: Brainstorming subject lines (Grade: B-)
I know, I said I don't use AI to write copy. I don't! But I have sometimes asked AI to suggest subject lines that will optimize open rate.
Three times, I tried subject lines suggested by AI:
But as a group, the AI subject lines had much worse open rates than my average.
Since the AI suggestions performed worse than the subject lines I wrote myself, I quickly stopped using them. But I sometimes still ask AI to write subject line suggestions, and I use these as inspiration to come up with something I like.
I've tried both general-purpose AI and marketing-specific tools for this use case. I don't find much difference in the quality.
#4: Ideas for data visualizations (Grade: C-)
I design all my own charts using Excel. I am not a designer, so I stick to simple patterns that I know will work. But every once in a while, I ask AI for advice on the best way to visualize a particular data set.
I've found the suggestions to be either obvious or obtuse, so I have yet to use any of them. In fairness, I haven't tried any purpose-built AI software for this use case; as with writing, making the charts helps me think through the data, so it isn't a big enough problem for me to have sought our purpose-built software.
#5: Brainstorming ideas for new topics (Grade: F)
Every once in a while, I have asked AI to play content strategist by suggesting new topics that nerd processor readers will enjoy. The suggestions are hilarious and terrible. For instance, here are some recent topic ideas suggested by AI:
- "The Power of Positive Thinking at Work"
- "How to Be More Productive in 2026"
- "Leadership Lessons From My Morning Routine"
Bland and generic, no data, no POV, and certainly nothing provocative. If a human content marketer made these suggestions, you would fire them immediately.
The bottom line: I don't use AI to write nerd processor, make my charts, or draw my comics, but I've used AI in a variety of supporting roles. As a content strategist, AI is ridiculously bad. But as a data analyzer, AI is an excellent sidekick.
Kieran
Flash sale! I am offering Viral Data Stories 101 (my course on telling data stories that go viral) at a 65% discount for ten subscribers this week, first come first served. A great gift for aspiring data storytellers in your life, including you!
My latest data stories | Build like a founder | nerdprocessor.com