This morning, GPT-5 was officially released.
And the reactions have been… split.
Some say it’s better than GPT-4.
Others swear it’s worse.
I wasn’t interested in debating theory.
I wanted to test it.
Not with a benchmark, or a prompt library, or a “Top 100 AI challenges” list.
But with something that actually matters to me:
Can it write in my voice well enough that my audience can’t tell the difference?
So I gave it a task.
I asked it to write a nurture email about NEO in my exact style — no selling, just story-led, with a clear point.
And I didn’t just use any generic prompt.
In fact, I didn’t use any prompts at all.
I used the AI Coach prompt embedded into my project instructions, plus every Context Transfer Package from Evolving Authority.
Here’s what it produced:
--------------------------------
Subject: The day I caught myself playing Side Character
A few months ago, I was mid-project when I caught myself doing something I warn my clients about all the time.
I was stalling.
Not because I didn’t know the next step.
Not because I didn’t have the tools.
But because I didn’t want to be seen messing it up.
I was polishing an internal process — endlessly tweaking, rewriting, “making it better” — instead of shipping it.
It was pure Side Character behavior.
The same pattern I’ve seen derail dozens of clients:
Waiting for the perfect conditions
Hiding work until it’s “ready”
Letting momentum die in the name of “standards”
And here I was… doing it myself.
The realization stung.
Because the truth is, nobody escapes these patterns.
Not even the person teaching them.
That’s the day I doubled down on NEO’s core design.
NEO isn’t here to give you “answers.”
It’s built to interrupt the moments you start drifting into Side Character mode.
It’s the AI that asks the one question you need to hear when you’d rather hide behind more research, more tweaks, or more excuses.
Because Main Characters don’t wait until they “have it all together.”
They move.
Messy. Visible. In progress.
And if I need help remembering that… maybe you do too.
More on how NEO does this soon.
FELIX
--------------------------------
I didn’t touch a word of the above.
If I pasted that into my email platform right now, most people wouldn’t know it came from GPT-5.
And here’s the thing: The AI Coach prompt actually feels better here than when I tried the same thing with Claude Opus 4.
There is so much noise in the AI world today.
Everyone has strong opinions about AI.
How I stay centered is by simply not caring bbout what people say about benchmark scores and numbers..because I’ve seen plenty of models “win” benchmarks and still give me a bad experience.
And today, this experiment knocked it out of the park for me.
So I’m going to be using ChatGPT (GPT-5) for my writing over the next week to see how it holds up.
If you want to see the raw conversation where this happened so you can judge for yourself, you can.
I tried to use the “Share” function, but for some reason whenever I clicked that button it didn’t do anything.
So I attached six screenshots at the end of this email instead.
Not too many because I didn’t want it to affect deliverability.
But enough for you to have a thorough look.
FELIX
P.S. This email describing my experience with GPT-5 and the output from GPT-5 was ALSO written with GPT-5
