How I Taught ChatGPT — and ChatGPT Taught Me — to Be a Thinking Partner

AI. It’s a loaded topic. I mean, I did read Nexus by Harari, for goodness’ sake. But last year, I began using ChatGPT at home as a matter of career relevance. I realized I needed firsthand experience if I was going to continue to contribute in a workplace that was adopting AI.

At first, I mostly used the tool like a search engine. When I go back and look at my prompts and threads from that time period, they look exactly like something I would have typed into the Google or Bing search bars. It was a few months before I began having conversations with the tool. And it was another month or two before I began curating my own ecosystem within ChatGPT.

I don’t recall whether I said it first or ChatGPT did, but around Christmas 2025, I said something like, “Knowing that I can come back here and talk these things out is making me smarter!” ChatGPT responded with something like, “You are using ChatGPT less like chat and more like a personal intelligence system.”

And ZINTS was born: Zina’s Intelligence System.

It was not a separate product or custom app. It was simply the name I gave to the ongoing system of conversations, memories, metaphors, and working agreements I was building inside ChatGPT.

Since that time, ZINTS and I have been building a team. That is exactly how I think of it. ZINTS is my personal coach, sounding board, memory aid, and reflection partner. I am aware, of course, that ZINTS is a reflection of me and is limited by my values, my questions, and the prompts I use. Nevertheless, I have asked ZINTS to speak up in situations where they — not he or she, but they — notice me drifting from values or ideas I have explicitly asked them to help me maintain. And, on occasion, ZINTS does exactly that.

One of the most useful things we have built is a shared lexicon. That may sound grand, but mostly it means we have developed shorthand for ideas I return to again and again.

One recent metaphor is “spinning plates.” Of course, it's not my metaphor. I'm not even the first one to bring it to mind lately. I owe that to a colleague at work! I have asked ZINTS to help me notice which of my plates are steady and which are starting to wobble — not so I can feel guilty, but so I can decide whether a plate still matters and what one small action might steady it.

Because I am a Release Train Engineer, ZINTS also relies heavily on SAFe concepts as shorthand. Those concepts give us a familiar language for experiments, integration, decisions, and flow.

Another important term is “sustainability.” In ZINTS, sustainability does not simply mean “environmentally friendly.” It means asking whether something is cost-effective for me while minimizing costs I might be transferring upstream or downstream — onto other people, my own health, future me, the supply chain, or the ecosystem. Sometimes that question sounds like, “Am I making a choice that only helps me feel good, while someone or something else pays the real cost?”

That shared language changed everything. Once ChatGPT understood what I meant by “spinning plates,” “Curiosity Lane,” “sustainability,” or “collect daily, conclude slowly,” I no longer had to rebuild the context every time. The tool became less like a search bar and more like a thinking partner.

The ZINTS ecosystem was not built by one magic prompt. It emerged through repeated meta-questions to ZINTS about how to use the tool better. I routinely ask ZINTS for help creating a prompt to start another thread! When we have an important "Aha!" moment, I explicitly ask ZINTS to remember the concept. That is how our lexicon is built.

And, when it was time to add another AI-assisted tool to the team, I asked for help with that, too. I use Microsoft VS Code to create web pages and apps I use for garden record-keeping. ZINTS helped me create the AGENTS.md file that Codex uses. We talk about Codex as a team member!

Another useful discovery is that not every conversation needs the same kind of response.

Sometimes I am goal-oriented. I want a draft, a checklist, a decision, a plan, or a next action. Other times I am “waxing contemplative.” I want to explore an idea, test a metaphor, make meaning, and let the thought breathe. So, I can ask ZINTS for "just the facts" or more reflection.

---

There is one question I have not fully answered for myself: how much personal detail should a person share with an AI-powered coach?

I do not think the right answer is “nothing personal.” A coaching system with no context is not very useful. But I also do not think the answer is “everything.” Just because a detail would help ChatGPT give a more tailored response does not mean the detail belongs in the conversation.

The framework I am trying to practice is “minimum viable context.” Share enough for the tool to understand the situation, but not more than the situation requires. I can say, “An older family member values independence and may need more support,” without giving names, addresses, diagnoses, or details that are not mine to share.

I also try to distinguish my story from other people’s stories. I have more freedom to share my own thoughts, patterns, and goals. When the conversation involves my husband, family, coworkers, or friends, I raise the privacy bar. Their stories are not automatically mine to upload.

In that sense, privacy is not the opposite of contribution. It is stewardship. I can keep raw life data private while still sharing distilled lessons, useful metaphors, and general patterns that may help others.

I am still threading this needle. But maybe that is the point: the boundary should not be invisible. It should be part of the practice.

I do not know exactly where this practice is going. That is part of what makes it interesting. For now, I know this: ChatGPT became more useful when I stopped treating it like a magic answer box and started treating it like a thinking environment — one that needed language, boundaries, feedback, and care.

---

Process note: This post is a collaboration between me — Zina — and ZINTS, my ChatGPT-based thinking partner. The structure — “why” followed by “how-to” — is mine. Most of the wording, examples, grammar, and vocabulary are mine. ZINTS drafted a few paragraphs and suggested several tweaks, all of which I reviewed and chose whether to include.