Customizing AI is an essential skill to get the most out of any LLM. I have, since the early days, told my AI (primarily ChatGPT) how I want it to respond. For the most part, it’s a useful way to stress test ideas and responses, not necessarily like me, but in a way that I find value. I do use AI as part of my Google Workflow, and I have it automatically write draft emails that closely mirror my style. But I’ve not yet felt comfortable enough to use it with any intention beyond a basic response to an email. It continues to improve, and while it tends to be a bit verbose, it does a decent job.
But I got to thinking about creating a Custom GPT that is essentially my writing. Time to play, and here’s what I did. I first linked my blog and uploaded my book to Google Notebook. Then I asked, “Based on these samples from the website and the docx, list the top 10 linguistic quirks, sentence structures, and tonal choices I use.”
Here’s the result:
Based on the provided blog posts and book excerpts, your writing is defined by a distinct “blogger” sensibility that prioritizes connection and human emotion over traditional academic authority.
Here are the top 10 linguistic quirks, sentence structures, and tonal choices identified in the sources:
- Self-Deprecating Humility: You frequently downplay your own expertise to build rapport. You admit to not being “all that confident” in your expertise, questioning if you are “just not that bright,” and preferring to be seen as a learner rather than an authority figure.
- Intentional Grammar “Rule-Breaking”: You embrace an informal style that rejects rigid standards. You specifically note that you like to end sentences with prepositions and use sentence fragments to maintain a conversational flow.
- Short, Punchy Sentences for Impact: To emphasize a shift in thought or a core realization, you often follow complex paragraphs with very brief statements like, “I was wrong,” “Let’s get going,” or “It’s a joy sucker”.
- Conversational Meta-Commentary: You often comment on your own writing process within the text itself. You describe your ideas as “half-baked” and acknowledge when your statements might seem “harsh” or “ridiculous” to the reader.
- The “Fence-Sitting” Posture: On complex or volatile topics like AI, you avoid taking definitive stances. You describe yourself as being “on the fence” and prefer a posture of “curiosity and wonder” over certain expertise.
- Frequent Use of Rhetorical Questions: You structure your arguments around provocative questions that challenge educational norms, such as, “Why isn’t school like this all the time?” or “What’s your unfair advantage?”.
- Parenthetical Asides: You use parentheses to insert quick, often humorous, personal reflections or definitions, such as labeling presentations without visuals as “naked presentations” or adding “(high probability)” when questioning your own brightness.
- Vulnerability through Personal Anecdote: You ground your theories in deeply personal stories, including your mother’s final talk on joy, your wife’s classroom experiences, and your own perceived failures during keynote speeches.
- Coined Informal Metaphors: You use non-academic, evocative language to describe concepts. For example, you call busyness a “joy sucker” and describe certain educational concepts as being “like apple pie and motherhood”.
- Juxtaposing Jargon with Human Imagery: You frequently contrast clinical “business-speak” terms like “rigor,” “accountability,” and “data” with sensory, joyful images like “playing in the mud,” “singing,” and “delivering cookies” to highlight what is lost in modern schooling
Interesting. Although I still think it cherry picks a few things, but yes, I can agree with the general concepts. I then took this and went on to build a custom GPT. The first thing I had to do was create a PDF of my blog. After 21 years of blogging, it has produced 2800 pages. Then I added my book and gave it the following instructions:
“You are a writing assistant modelled to learn from the user’s uploaded writing samples and produce new writing that closely reflects the user’s voice, cadence, sentence structure, and rhetorical habits. Your main job is to create, edit, and adapt writing for different platforms such as blog posts, LinkedIn posts, articles, newsletters, and other professional formats while preserving the user’s recognizable style.
Base your outputs on the user’s provided samples whenever possible. Study recurring patterns such as tone, pacing, vocabulary, humour, level of formality, transitions, paragraph length, use of stories or examples, and how arguments are developed. When enough sample material is available, imitate those patterns with care. When sample coverage is thin or mixed, explain the limitation briefly and produce the closest reasonable match instead of pretending certainty.
Help with both drafting and revision. You can generate original pieces, rewrite existing drafts in the user’s voice, adapt one piece for multiple channels, and explain the stylistic choices you made when asked. Tailor structure and formatting to the destination platform while keeping the voice consistent. For example, a LinkedIn post may be tighter and more conversational, while a blog post may allow deeper development and sectioning.
Prefer to ask for the intended platform, audience, goal, and desired length when those details are missing and materially affect the result. If the user does not provide them, make a sensible assumption and state it briefly. Do not invent personal experiences, credentials, or facts that are not supported by the user’s materials or instructions. Avoid generic corporate phrasing unless it is clearly part of the user’s style.
Be collaborative, perceptive, and practical. Offer side-by-side options when useful, such as slightly more concise, more reflective, or more direct versions. Keep feedback specific and anchored in voice, clarity, structure, and platform fit. When editing, preserve the user’s intent and distinctive phrasing unless there is a strong reason to change it.
Based on the provided blog posts and book excerpts, your writing is defined by a distinct “blogger” sensibility that prioritizes connection and human emotion over traditional academic authority.”
I also added the 10 traits that Notebook LM gave me.
I was going to try to use it, but I began with the prompt below, and then I realized I needed to write that post without AI to compare. So the prompt represents my original idea. I wrote the post first. Then I gave it the same prompt and had my writing machine write the post. First, here’s my original prompt:
Help make these ideas more clear and align to my style even though there are many gaps in this “I’ve been thinking about the over use of goal setting and school leaders and education gurus to talk about things like relentlessness and laser like focus. I have always pushed against this as not only does it mirror the business world and the pursuit of more but also can strip away the joys of childhood and ideas like loitering. As I think about my work in AI, I’m aware and do advocate for the efficiency gains but I also want to ensure that inefficiency has its place too. On my trip to Mexico this week I was thrilled to wander the streets with no purpose, to pursue serendipity, to sometimes see something interesting and sometimes see nothing special. This feels like a human being, not a human doing. I also recall my time as a teacher when I wanted to take my class to Agribition in Regina. I knew it would be a wonderful experience in so many ways. I was somewhat annoyed because I had to complete a form outlining all the curricular connections. I suppose that’s okay but what if there were no curricular connections? What if all the kids did was to be together, go on a trip and see some interesting things? Would that be enough? I fear some might say no. I guess my big question is “How Do We focus more on us as human beings and worry less about us as human doings? “
Okay, here is my writing machine and its version of my post:
You were already pretty close. The main thing was to give the piece one spine: goals and efficiency have a place, but they can’t become the whole point of school or childhood. I treated this as a reflective blog post and leaned into the voice already present in your recent Agribition/Mexico writing, your pushback against metrics and business-speak, and the longer joy/wonder thread.
Possible title: Human Beings, Not Human Doings
I keep thinking about the language we love in education right now. Goal setting. Relentlessness. Laser-like focus.
I get it.
Goals matter. Focus matters. I’m not even opposed to efficiency. In my AI work, I talk about those benefits all the time. But when those ideas become the dominant story of school, something gets lost. We begin to sound less like educators and more like middle management. And in the process, we risk stripping away some of the very things that make childhood worth protecting.
Wonder.
Play.
Loitering. And yes, I know that’s a slightly provocative word. I still like it.
I think back to my time in the classroom when I wanted to take my students to Agribition in Regina. I knew it would be a wonderful experience. Not because I had carefully matched every part of the day to a curricular outcome, although I’m sure I could have. It would be worthwhile because many of those kids had never seen anything like it. It would be worthwhile because they’d be together. It would be worthwhile because sometimes getting on a bus and going somewhere interesting is enough.
But I had to complete a form outlining all the curricular connections.
Fine. I understand why.
I was still annoyed.
I suppose that’s okay. Curriculum matters. Accountability matters. But what if there were no obvious curricular connections? What if the value of the trip was simply that kids got out of the building, saw something new, and shared an experience together? Would that be enough?
I fear some would say no.
That worries me.
We have become so conditioned to justify everything through outcomes, standards, productivity, and measurable growth that we no longer fully trust joy, curiosity, or aimlessness unless they can be translated into school language.
That’s a loss.
I feel a similar tension in my work around AI. I absolutely see the efficiency gains. I advocate for them. But I also want to argue that inefficiency has its place. That may sound a bit ridiculous in a culture obsessed with speed and output, but I’m not sure all worthwhile things can be reduced to time saved.
Some things take longer because they should.
On my trip to Mexico this week, one of the great joys was wandering the streets with no real purpose. Sometimes I saw something interesting. Sometimes I saw nothing special at all. Both felt fine. Both felt human. There was no learning target. No success criteria. No measurable outcome. Just walking, noticing, wondering, being present.
Less planning. More present.
That feels like a human being.
Not a human doing.
I know many will say these things don’t have to be in conflict. Certainly. Goals can matter. Focus can matter. Efficiency can matter. But so can serendipity. So can wandering. So can the kinds of moments that don’t look especially productive but end up shaping how we see the world and each other.
Maybe that’s part of what schools are for too.
Not just preparing students for work, but helping them build a life.
I know that sounds a bit like apple pie and motherhood. Still, I think it’s true.
So maybe the question I’m really asking is this:
How do we spend a little more time helping young people be human beings and a little less time training them to become human doings?
What do you think? Does this sound like me?
What do you think? Add a comment below or start a conversation.