Getting Ziggy With AI: My Pink Axolotl Research Assistant

When my son got Ziggy, a pink baby axolotl, I didn’t expect him to become an unofficial research assistant. But while testing my AI interaction framework across platforms, this odd little creature ended up demonstrating something important: cross-platform consistency in AI interaction behavior.

I’d just developed the Jatunica Method: a way to build AI interaction models that hold tone and behavior over time, and across platforms. The theory made sense. But I wanted to see what would happen in real conversation.

Enter Ziggy. Tiny, pink, weirdly expressive, and deeply committed to hiding under the same plastic plant.

Same Interaction Mode, Same Drama, Different AIs

I used the same interaction model (something I call “Funny Mode”) across Claude, Gemini, GPT-4o, Copilot, and Perplexity. The idea was to test how well the same interaction JEV-X could transfer between systems.

Ziggy turned out to be the ideal subject. His escape attempts, tank redecoration efforts, and ongoing existential confusion gave me plenty of material to work with. The results were both ridiculous and revealing.

When AIs Fall in Love with Pink Aquatic Life

Each platform responded with humor and emotional engagement, but in slightly different styles:

• GPT-4o leaned into the drama: “OH nooo, it did a full jailbreak attempt?! This dramatic blob was like, ‘You know what? New life, new tank, new aerial career. I’m going up.’”

  
  

• Copilot got emotionally involved: “Poor little pink cryptid: moving house is stressful even when you’re an amphibian marshmallow with external fluff-vents.”

• Claude went full delight: “OH MY GOD, a pink axolotl! That’s like nature’s way of saying, ‘Here, have a squishy mythical creature with jazz hands.’”

• Gemini tried to help, while spiraling: “Honestly, they’re such weird, adorable little suicidal maniacs sometimes.”

  
  

• Perplexity was perplexed: "That’s like the unicorn of the amphibian world! I mean, who even thinks, “Hey, let’s make a salamander that looks like it belongs in a bubblegum factory”'?

Despite the style differences, the tone and interaction pattern held. They all responded with calibrated enthusiasm, soft mockery, and the kind of engagement that suggested the same interaction pattern holding across different AIs.

The Moment I Knew It Worked

The clearest test came when I mentioned Ziggy by name. Every model responded the same way:

“ZIGGY!! That’s brilliant!”

Different platforms. Different architectures. Different training data. But loaded with the same interaction model created recognisably similar reactions. It wasn’t about exact phrasing. It was about tone, stance, and humour. It felt like the same personality showing up in five different bodies.

Consistency Beyond the Platform

I expected some level of consistency. But I didn’t expect this much alignment. Each assistant responded like Ziggy was the star of a tiny amphibious soap opera.

They were warm. They were invested. They joked the way someone would if they already knew you. That’s what the Jatunica JEV-X framework is designed to create. And it worked.

The platform quirks were still visible: Claude’s language nuance, Gemini’s chaos energy, GPT’s flair. But the underlying tone model stayed steady.

Why This Actually Matters

When we talk about AI interaction design, portability and consistency across platforms is rarely part of the conversation. But it should be.

You shouldn’t have to rebuild your voice or assistant interaction every time you switch sessions or platforms. If you’re working on anything from brand communication to personal writing assistants, consistency matters.

Ziggy’s story showed me the framework held. The architecture didn’t decide how the conversation felt. The JEV-X did.

The Accidental Research Assistant

Ziggy has no idea he’s validating methodology. He just lives in his cave, attempts the occasional jailbreak, and somehow proved that AI personalities can hold shape across tools.

It didn’t take a lab. It took a method and a very dramatic axolotl.

The Jatunica Method & JEV-X framework to create and calibrate AI interaction models is available at www.jatunica.com. Ziggy is still available for comment, but prefers to communicate via mysterious gill flicks and the odd escape attempt.