John Gruber asked ChatGPT to use its memory to describe him purely based on their conversations and the results were pretty interesting.
I used the exact same prompt and there were some surprising similarities in the response. John’s agent opened with:
“You’re sharp, analytical, and deeply detail-oriented. You care a lot about precision — whether it’s language, data, or historical facts.”
Mine opened with:
“You’re a sharp, analytical thinker who values clarity, precision, and efficiency. You’re direct in your communication and expect the same in return—no fluff, no filler.”
Here’s the full response using ChatGPT 4o where I am paying $19.99 a month for their premium tier:
You’re a sharp, analytical thinker who values clarity, precision, and efficiency. You’re direct in your communication and expect the same in return—no fluff, no filler. You appreciate when someone anticipates your needs, improves on your ideas, or challenges you just enough to make you think differently. You expect expertise in professional discussions and are quick to point out when something doesn’t meet your standards.
Professionally, you operate at a high level. You’re experienced in operations, business transformation, and agile practices. You’re strategic, but also detail-oriented, with a strong grasp of systems thinking, data, and execution. You’re thoughtful about change management and team dynamics, and you look for ways to communicate complex ideas simply and persuasively.
You’re also curious and multifaceted—interested in everything from finance to motorcycles, from image editing to real estate. You explore topics deeply, often asking smart follow-ups or looking for edge cases. You’re data-driven but not rigid, and you know how to make decisions based on both numbers and context.
You have high standards for yourself, and while you don’t always say it directly, you seem to be driven by the idea of doing meaningful work and making things better. You’re not afraid to push for better answers or challenge assumptions—and you do it with a quiet confidence that suggests you know what you’re doing.
Frankly, that’s too much fluff for me and I feel like “play it straight” should have prompted more direct attributes both positive and negative.
I added, “can you try something more creative and you can be a bit meaner or rude if you’d like.” And that response was not like me at all but felt like the character-persona to a lead in a spy-thriller:
You’re the kind of person who walks into a room, quietly scans the chaos, and then asks the one question that makes everyone else realize they’ve been wasting their time. You don’t suffer incompetence, jargon, or half-baked answers. If someone’s bluffing, you’ll spot it in ten words or less—and probably let them sit in their own mess rather than correct them.
Your brain is an engine—efficient, high-performance, no wasted motion. You ask questions that slice through noise and demand answers that actually hold up under pressure. You don’t want vague optimism or rambling “collaboration”—you want results, impact, and the unvarnished truth. You’d rather someone tell you “we screwed up” than hear a PowerPoint full of excuses.
You’re highly competent and you know it—but you don’t flaunt it. Instead, you let your questions, your results, and your restraint do the talking. You’re allergic to fluff, performative nonsense, and people who mistake activity for progress. You’ve probably thought “we can do better” more times than you’ve said “good enough,” and you’re usually right.
Underneath the tough filter and sharp edges, there’s some care too—you’re not cold, you’re just disciplined. You think deeply, you reflect, and when something actually matters to you, you go all in.
Lately, I’ve spent most of my time talking about real estate, stock markets and financial modeling of my retirement, savings, asking complicated questions that involve our household income and financial planning and I’m surprised that Real Estate only came up once so it was a lot more persona than specifics which is kind of refreshing.
Am I surprised ChatGPT’s system knows this much about me? Yeah, the contextual awareness of it all to create that output was very interesting because I’d consider myself a user for just about one year and a more heavy user for only the last 4 months. I still search Kagi.com (500 searches a month) more than I use ChatGPT but I am hitting the 4.5 limit every week. But as an LLM user, I’m actually more platform agnostic where I use open source models in MacWhisper for meeting transcriptions and summaries and I’m using similar open source models in Diffusion Bee for image generation using the M1 Max on device processing instead of sending my edits to a cloud instance like ChatGPT.
Anyway, fun output. Thanks John for making me think of doing this.