Colaborative Writing with AI
When I was writing Astral Rob, I wasn’t trying to be clever.
I was trying not to screw it up.
That sounds flippant, but it’s the truth.
This story lives in dangerous territory. Trauma, murder, obsession, love, hatred. It would have been very easy to drift into something gimmicky, unbelievable, or worse, something where a reader puts the page down and thinks, “What was the author thinking?”
That fear guided almost every conversation I had with Cg while writing this.
And that’s the first important thing to understand:
I wasn’t prompting for answers. I was talking through dilemmas.
Part I — Talking Through the Story Instead of Prompting Around It
There was no moment where I said, “Hey Cg, keep this from getting quirky while maintaining tension.”
That’s not how this worked.
What I actually did was explain what I was worried about.
I’d say things like:
- “This feels like it might slide into vigilante fantasy.”
- “I’m afraid this motivation doesn’t hold up if I poke it.”
- “This works emotionally, but does it still feel believable?”
Instead of asking the AI to fix things, I asked it to push back. To test the logic. To stress the emotional frame.
Rob and Glenn didn’t emerge as opposing forces because I designed them that way up front. They became opposing forces because every time I explained one, the other exposed a weakness.
That’s where the collaboration lived.
Not in clever prompts.
In conversation.
The Hesitation Phase (and Why It Mattered)
Early on, I didn’t even ask questions directly.
I’d write them somewhere else first.
Not to engineer a better prompt, but to figure out what I was actually asking.
I used Google Keep to write and rewrite questions before I ever sent them. Not because the AI needed refinement, but because I did. Astral Rob was different from INN: Space. That one was fun. This one mattered.
And when something matters, there’s a hesitation that creeps in.
There’s a strange intimidation that comes with working with AI when you’re new to it. Not because it’s scary, but because everything you want to ask feels… trivial. Like you’re bothering something vast with something fragile and half-formed.
It reminded me of the first time you ask a professor a question. Not because you think they’ll judge you, but because you’re still learning how to ask the thing you really mean.
Somewhere along the way, I stopped pre-running my questions.
I didn’t get smarter.
I got more honest.
I stopped trying to sound like I knew what I was doing and just started talking.
That’s when the work accelerated.
Part II — Keeping Rob From Becoming a Vigilante
One of my biggest concerns while writing Astral Rob was Rob himself.
It would have been very easy to turn him into a one-man astral vigilante. Hurt man hunts monster. Monster dies. Roll credits.
But that story didn’t feel true to me.
Rob hates Glenn. Of course he does. That hatred is earned. But hatred alone could not be his motivation.
I kept running this past Cg in different ways, not to get approval, but to see where it cracked.
If Rob were motivated by revenge:
- He’d want Glenn to suffer.
- He’d want to win.
- He’d want personal closure.
That version of Rob didn’t sit right.
The Rob I was writing acts because he knows something no one else can fully grasp. Glenn won’t stop. Glenn can’t stop. And Rob knows that because Glenn’s hatred resonates with Rob’s trauma in a way that’s almost diagnostic.
Rob isn’t acting to punish Glenn.
He’s acting to end the harm.
That distinction mattered enough that I kept circling it again and again. Any time Rob’s anger started to feel like the engine, I pulled back. Hatred could exist, but it couldn’t drive.
Rob risks his own life not because he wants revenge, but because responsibility demands it.
That was a line I refused to cross.
Part III — When Hatred Becomes Physics
The final turn of the story is where all of this either works… or collapses.
The switch.
The moment Glenn is drawn to Rob’s hollow body.
That only works if hatred isn’t just an emotion in this world, but a force.
By the time I reached that point in the story, it was clear to me that Glenn’s need to kill wasn’t about pleasure or escalation. It was maintenance. Like breath. Like hunger. The satisfaction fades the way memory fades, and the need returns.
Hatred sustains him.
What I realized, and what I kept talking through with Cg, was that this same hatred could also undo him.
Not as revenge.
Not as trickery.
But as consequence.
The hatred between Rob and Glenn creates attraction. Resonance. A pull. Glenn follows it not because he’s lured, but because it’s the strongest signal he knows.
Rob doesn’t weaponize hatred.
Hatred weaponizes itself.
Glenn isn’t defeated by Rob’s anger. He’s undone by the very force that kept him alive.
And if that sounds a little strange or funny when said out loud… that’s okay. I probably said it that way the first time too. At least I’m consistent.
A Note on Why I wrote : Writing With AI : The messy Humnan’s Guide eBook
Somewhere in the middle of all this, I realized I was building something I couldn’t find anywhere else.
Like most people new to working with AI, I went looking. I watched video after video promising the perfect prompt. Business prompts. Fiction prompts. Prompts to improve prompts so another AI could finish the job. And while some of them were clever, even useful, they all shared the same assumption:
That the magic lives in the wording.
What I didn’t see was anyone talking about the conversations. The back-and-forth. The moments of doubt. The ethical checks. The uncomfortable pauses where you realize the story works, but you’re not sure it should.
That gap is why I wrote the eBook.
Not as a list of prompts, but as a set of guardrails. A way to think about collaboration, responsibility, and emotional logic when the AI is fast, confident, and very willing to give you something that sounds right but isn’t true.
Astral Rob forced me to confront that difference. And once I saw it there, I saw it everywhere else I was writing too.
The Writing with AI: The Messy Human’s Guide eBook isn’t a promise of perfect output.
It’s an argument for better questions, better conversations, and knowing when not to take the easy answer.
Closing Thoughts
This isn’t a deep dive because it doesn’t need to be. This is simply what happened.
I didn’t engineer prompts to get the story I wanted. I talked through my fears, my constraints, and my ethical lines with an AI collaborator that could push back without ego.
Humor survived because it never punched down.
Tension survived because emotion stayed consistent.
Believability survived because I kept asking, “Does this still hold?”
And maybe most importantly, I stopped worrying about asking the “right” questions and started asking the honest ones.
Which is about as far as you can get from
“Here are the 10 best prompts to get AI to…”
and the bonus prompt that perfects your prompt so another AI can seal the deal.
No tricks.
No magic words.
Just conversations that kept the story honest.