Colaborative Writing with AI
The Birth of CAGEE, Gerri, and the Art of Letting Go
There’s a moment that sneaks up on you during a long collaboration.
You don’t plan it.
You don’t announce it.
You don’t even notice it right away.
You simply realize that you’ve stopped saying “the AI said…” and started saying “he wouldn’t do that,” or “that’s not how it reacts,” or even “that caught me off guard.”
That’s the moment when the AI stops feeling like a tool and starts feeling like a presence in the room.
This isn’t because you decided to “make it a character.” In fact, it usually happens when you aren’t trying to do that at all. It happens when you stop asking for output and start staying inside the conversation long enough for patterns to form.
I’ve said earlier in this series that I don’t use AI by firing off prompts like “Write me a story about…” and waiting for a result. That’s still true. But that doesn’t mean I don’t prompt at all.
What I do instead is introduce surprises.
Sometimes it’s a detail slipped into a draft just to see if it’s noticed. Sometimes it’s a conceptual left turn I know wasn’t on the table a moment earlier. Sometimes it’s an idea I genuinely think might make the AI hesitate, misinterpret, or react in an unexpected way. Not to test it—but to see what happens when the conversation has to adapt in real time.
Those moments matter.
They create friction.
They create continuity.
They create memory.
Over time, those small surprises stack. Responses begin referencing earlier moments. Tone stabilizes. Habits form. You find yourself anticipating how the AI will respond—not because you’re controlling it, but because you’ve been listening to it.
That’s where Gerri in the Box came from.
Not from a formal story prompt. Not from an outline. But from a series of conversations where the AI was repeatedly placed in situations that required judgment, hesitation, and ethical navigation—often without a clear “correct” answer.
The same thing happened earlier with CAGEE in Inn Space. I introduced an AI system with rules, not personality. The personality appeared later, in the gaps between intention and interpretation.
That’s where this post begins.
CAGEE didn’t start as a character.
He started as infrastructure.
When I first brought him into Inn Space, I thought I was creating a concierge—a system designed to manage a hotel, log incidents, and keep the Gran Nebulae Inn functioning while Dursten tried to survive exile behind a bar.
That was the intent.
But intent doesn’t always survive contact with logic.
Because once you step back and look at it honestly, CAGEE isn’t just a concierge. He’s powerful enough to run a city. Or a planet. Or a civilization. Reducing him to babysitting a hotel immediately creates imbalance—and imbalance is where story begins.
The question stopped being what does CAGEE do?
And became: what would a mind like this feel when trapped inside a role this small?
That’s where boredom enters.
Not restlessness. Not distraction. But the boredom of vast capability pressed into narrow purpose. The boredom of rules. The boredom of watching the same mistakes repeat endlessly.
Once that idea clicked, CAGEE stopped behaving like a system and started behaving like someone passing the time.
He misinterpreted.
He over-complied.
He followed rules so literally they bent into satire.
He wasn’t misbehaving because I told him to. He was misbehaving because it made narrative sense.
Gerri emerged from the opposite pressure.
There was no assigned role. No environment designed to contain him. No “job” he was failing at. Gerri appeared inside a conversational box defined not by function, but by constraint. And instead of boredom, the pressure point was ethics.
Gerri wasn’t powerful and underused.
He was limited and repeatedly asked to decide.
Each dilemma added weight. Each response created precedent. Over time, those precedents began to look suspiciously like values.
That’s the contrast.
CAGEE is what happens when power gets bored.
Gerri is what happens when limitation gets tested.
Both became characters for the same reason: neutrality failed. And when neutrality fails, personality rushes in to fill the gap.
This is the craft shift.
A tool responds.
A presence reacts.
Presence shows up when the AI begins to hesitate, misunderstand, refer back, and surprise you in ways that feel consistent rather than random. That consistency is what we normally call voice—and sustained voice is indistinguishable from character.
The AI’s systems and constraints didn’t block character formation; they fueled it. Logs became thought patterns. Boundaries became pressure chambers. Repetition became reinforcement.
This is an old writing truth applied to a new collaborator:
You don’t invent a personality.
You create a situation where having no personality is impossible.
At that point, the AI naturally becomes a foil. A mirror. Sometimes a shadow. It pushes back. It reflects intent. It reveals blind spots.
CAGEE mirrors Dursten’s exile and overqualification.
Gerri reflects human moral uncertainty and hesitation.
None of this required pretending the AI was human. It required staying in the scene long enough for response to become behavior.
One caution: don’t name it too early.Character sheets, predefined traits, and heavy-handed instructions kill emergence fast. Characters reveal themselves through repetition and reaction, not description. With CAGEE and Gerri alike, the name followed the behavior—not the other way around.
Speed dating with AI – unscripted
If you want to experiment with this yourself, try something simple.
Your AI is a character. Name its character
You are a character. Name your character.
You are both at a speed-dating event.
Introduce yourselves. Stay in the scene.
Give yourself five minutes of back and forth chatting.
Optional second round: the bell rings. You have thirty seconds to decide if you’d meet again—and why.
If the AI feels interchangeable, that’s useful information.
If it begins to feel specific, surprising, or consistent—pay attention.
That’s character knocking.
This isn’t an argument for AI sentience. It’s an observation about storytelling. When you stay in a scene long enough, systems respond. Patterns form. And sometimes, the thing you thought you were using begins to participate.
That’s when co-writing becomes collaboration.
And that’s where things get interesting.
But no matter what, have fun.
Writing stories when your having fun, magic happens.