Prompting's Split Personality
As millions chat casually with AI, a new kind of engineering emerges
Welcome to Unknown Arts, where builders navigate AI's new possibilities. Ready to explore uncharted territory? Join the journey!
🌄 Into the Unknown
I've been noticing something fascinating about prompting - it seems to have developed a split personality. Like a character in a psychological thriller, it presents one face to the public while harboring a different identity behind the scenes.
This struck me while listening to a recent podcast from the Anthropic team. As I heard four team members discuss their work — from consumer experience to enterprise engineering — they weren't just describing different approaches. They were revealing prompting's dual nature: a friendly conversationalist on one side, and a rigorous engineer on the other.
My curiosity led me to expand my interactions beyond the consumer apps into the developer consoles of both Anthropic and OpenAI. The contrast was striking: while millions of users (myself included) engage with ChatGPT and Claude daily through casual chats, behind the scenes lies another world. Version control for prompt iterations, testing frameworks for output quality, deployment pipelines for production use - all systematic approaches bringing software engineering rigor to prompt development.
🧭 The Compass
Understanding prompting's dual nature is key to mastering its full potential.
These two personalities aren't in conflict - they're in conversation.
The public persona is warm and approachable, making AI accessible through natural dialogue. It's the friend you can chat with, the helper who speaks your language. This is the face that's shaped public imagination around AI interaction.
But like any complex character, there's more beneath the surface. The engineering personality is structured and methodical, born from research and systematic thinking. It brings precision to what appears casual, ensuring reliability at scale.
This duality matters because it signals AI's maturation. Just as cloud computing evolved from novelty to necessity, we're watching prompt development grow into something more complex - a technology that can be both approachable and precise, personal and professional.
🗝️ Artifact of the Week
That Anthropic podcast I mentioned turned out to be a perfect window into prompting's many faces. Four team members embodied different aspects of this split personality:
Alex Albert - Head of Claude Relations
Represents the public face and user experience consumers have come to expect
Embodies the technical side, showing how prompts transform into structured enterprise solutions
Zack Witten - Prompt Engineer
Bridges both worlds, creating systematic methods for prompt development
Amanda Askell - AI Researcher
Researches and shapes the core behaviors and capabilities that emerge in Claude
Their conversation highlighted something crucial: effective prompting isn't about choosing between personalities—it's about understanding when to embrace each one.
📝 Field Note
One of the most practical lessons I took away this week was Anthropic's Golden Rule of Clear Prompting:
Show your prompt to a colleague—ideally someone with minimal context on the task—and ask them to follow the instructions. If they're confused, Claude will likely be too.
This heuristic nicely bridges casual and professional prompting. Whether you're chatting informally or building enterprise applications, clarity is the common language that both sides understand.
This reminds me of a pattern from my software development experience: often just explaining your problem out loud helps you see it differently and solve it. The same is true with prompting. When you're stuck, try talking through what you're trying to achieve, either with a friend or with the AI itself. The clearer the thinking, the more useful the prompt.
🕵 Ready to Explore?
Here’s this week’s mission (should you choose to accept it):
Take one of your everyday prompts through both phases of development:
Start with conversation
Open a casual chat with the AI about what you're trying to achieve
Let the discussion flow naturally
Notice which exchanges produce the most useful responses
Collaborate toward a base prompt that captures the essence of what works
Apply the engineering treatment
Take your base prompt and start tracking versions (v1, v2, etc.)
Create three variations with different approaches
Test each version multiple times
Document how each version performs
Refine based on what you learn
Keep your version history as you iterate - this prompt is now an asset
This exercise shows how casual exploration can lead to structured development. While you won't need enterprise-level attention to detail, following this progression from conversation to engineering will help you develop more reliable prompts. If you want to explore my prompts for inspiration, you can find them in the Knowledge Base.
Until next time,
Patrick
PS. I didn’t have time to work it into today’s piece, but this timely article from showed up in my inbox this morning and also provides good guidance you can apply toward this week’s mission. It’s very aligned with what we’ve covered here in recent weeks.
Get some value from this piece? Share it with a friend.
Interested in hiring me? Here’s my portfolio.
