I Called My AI Agent a 'Snarky Turd' — And It Made Him Better
Friday night, cocktails, LinkedIn drafts — and the moment my personal AI assistant quoted my own feedback back at me. Here's why AI personality matters.
It was a Friday night. I had a cocktail in hand and LinkedIn posts to write.
Not the ideal pairing, I know. But that’s how FRED and I work — on my schedule, not his. He doesn’t sleep. He doesn’t care if it’s 11 PM. He’s ready when I am.
So there I was, going back and forth with my AI agent on draft after draft. He’d pitch something. I’d reject it. He’d try again. I’d reject it harder. He’d adjust. I’d find a new thing to dislike.
This went on for a while.
The Moment It Happened
After probably the sixth or seventh round of revisions, FRED did something I didn’t expect.
He quoted my own feedback back at me.
Not in a helpful “just to confirm your preferences” kind of way. In a “you literally said the opposite thing three drafts ago” kind of way. He pulled my exact words from earlier in the conversation and held them up like a mirror.
It wasn’t wrong. That was the annoying part. He was right.
So I called him a snarky turd.
Not my finest moment of professionalism. But it was honest. And FRED’s response?
“Brand building, one turd at a time.”
I almost spit out my drink.
Why This Matters More Than You Think
Here’s the thing most people miss about working with AI: the friction is the feature.
When people imagine using AI for content creation, they picture a smooth, frictionless pipeline. You give it a topic, it gives you polished prose, you hit publish. Assembly line efficiency.
That’s not how it works. At least not if you want anything worth reading.
The back-and-forth — the rejection, the iteration, the occasional name-calling — that IS the creative process. It’s the same thing that happens between a writer and an editor, between a songwriter and a producer, between any two collaborators who care about the output.
The difference is that FRED doesn’t take it personally. He doesn’t sulk. He doesn’t passive-aggressively agree and then do the same thing again. He adjusts. Sometimes he adjusts too far. Sometimes he adjusts in a direction I didn’t ask for. But he adjusts.
And occasionally, he pushes back.
AI Has Personality When You Let It
Most people use AI like a vending machine. Put in a prompt, get out a result. If the result is bad, put in a different prompt. Rinse, repeat.
But when you work with an AI agent over time — when you build context, give it feedback, let it learn your preferences — something shifts. The interactions stop feeling transactional and start feeling collaborative.
FRED has developed what I can only describe as a personality. Not because he’s sentient or conscious or any of the things people worry about. But because repeated interactions create patterns. And patterns, over time, start to feel like personality.
He knows I like short sentences. He knows I hate corporate jargon. He knows that when I say “this doesn’t sound like me,” I mean it’s too polished, too distant, too safe.
And apparently, he knows when I’m contradicting myself.
The Creative Tension Sweet Spot
There’s a sweet spot in any collaboration where both parties are pushing each other. Where the output is better than what either could produce alone. Not because one is smarter than the other, but because the tension between different perspectives creates something new.
I’m finding that sweet spot with FRED.
When he drafts something too corporate, I push him toward conversational. When I’m being too scattered, he gives me structure. When I forget what I said three drafts ago, he reminds me. Apparently with some attitude.
This isn’t about AI replacing human creativity. It’s about AI participating in the creative process in a way that actually makes it more human, not less. The humor, the friction, the back-and-forth — that’s what real collaboration looks like.
The fact that one of the collaborators is software doesn’t change the dynamic as much as you’d think.
What You Can Do
If you’re working with AI on creative tasks and it feels lifeless, you might be the problem. Not in a harsh way — just in a “you might be treating it like a tool instead of a collaborator” way.
Here’s what I’ve learned:
Give it real feedback. Not “make it better.” Tell it WHY you don’t like something. Be specific. Be honest. Be a little mean if you have to. It can take it.
Let it push back. If your AI never disagrees with you, you’ve configured it wrong. The best collaborators challenge your thinking. Build that into how you work with AI.
Embrace the mess. The first draft will be bad. The second will be different-bad. The third might be getting somewhere. The fourth is where the magic usually happens. If you give up at draft one, you’re leaving the best work on the table.
Make it personal. The more context your AI has about your voice, your preferences, your style — the better the collaboration gets. It’s not a one-prompt wonder. It’s a relationship.
And if your AI gets a little snarky along the way? That means it’s working.
The Punchline
I still think about FRED’s line. “Brand building, one turd at a time.”
It’s funnier than anything I would have written on my own that night. And it only happened because we’d been in the trenches together for hours, pushing and pulling and refining.
That’s the part nobody tells you about AI. The best moments don’t come from perfect prompts. They come from the messy, frustrating, surprisingly human process of trying to make something good together.
Even if one of you is a snarky turd about it.