How I Use AI to Think, Diagnose, and Communicate as a Team of One
How I Use AI to Think, Diagnose, and Communicate as a Team of One
AI as my strategic co-pilot, not a shortcut but a thought partner.
Context: Leading Without a Playbook
In APAC, my role leading Project Management often felt like being the founder of a start-up inside a large company.
- Few usable systems or templates, and imported frameworks often felt irrelevant after a couple presentations.
- Infrastructure fragmented across markets.
- Teams who had never seen project management as anything more than scheduler.
- And, in China, a VPN bubble that cut me off from most AI tools. I joked I was “building the plane while flying it.” That was true, but the bigger challenge was that teams could not even agree on whether to take a plane, a train, a car ride, or simply walk. In reality, we were doing all of those at once, in different directions, with no shared map. That multi-dimensional “chaos” was a gap I could not concretely understand until I started using AI tools. Massive credit to YouTuber Vicky Zhao who taught this approach and I immediately understood its power of using AI to help me define the problems better so I can solve it, not the other way around. Here’s the video that unlocked so much for me.
The Mindset Shift: AI as a Thinking Partner
I don’t treat AI as a crutch. While I give my AI the role of my board of advisors, I also approach it with caution and ensure I understood its sources and inference. My borrowed simple 3-question framework to check my thinking partner (the AI):
- What assumptions are you (AI) making in your recommendations/answers?
- What are the root insights? Use the Five Why Model against those assumptions.
- What alternatives are there? This 3-step process reveals to me what assumptions AI is using in its response to me, and allows me to gain perspectives that were hidden from me or call me out when I’m overthinking.
The Tools I Reach For
I use a stack of tools, yes it’s me Mr. Gearhead and his tools, but each do play a unique role that I’ve found helpful:
| Tool | Use |
|---|---|
| Elicit | For surfacing expert references and insights without endless Googling - ranked by most cited. No, I don’t read them in depth. Checkout the video for how I digest it. |
| NotebookLM | My document cross-analyzer/integrator of patterns and themes, where cited research, my material and memos talk to each other. I extract outlines that’s overly detailed and cited - just my style. |
| ChatGPT | My main thinking and articulation partner, helping me make those detailed outlines frame ideas and a-ha moments, then pare them down into decks and comms for different audiences |
| Claude & Gemini | Fresh eyes that don’t know my personality and history, which helps me spot repeating patterns. It’s my “second opinion” when I feel overly influenced by ChatGPT. The results end up usually in very similar areas. |
| Copilot | Occasional context/person aware message editor internally, underutilized back in APAC days, but currently the important interface in my AI OS process currently in the US. (GPT5 helped kick it up a notch as of two weeks ago!) |
It’s not about the tools themselves. It’s about creating a rhythm of perspective, compression, and clarity. Use the tools that can help reveal them for you! And maybe share out your steps for others too. Sharing is caring.
One Example: Diagnosing a Meta-Problem
Suppose I need to explain an operational issue to a regional COO. On the surface it looks like missed deadlines or insufficient headcount. But the deeper problem is misalignment between specialty teams across markets - something I sensed but could not articulate nor be confident about. Could it be a wider systematic challenge or “I am the problem”? Here’s what my process looks like:
- NotebookLM houses various research papers on mitigating friction in multicultural virtual organizations and teams to help provide perspectives and context for me, followed by refining the problem I’m describing through voice dictated memos I have (and abundant color coded sticky notes on shelf doors and walls for me to describe what I’ve noticed). I gain some initial framework language and repeated insights from others to ground me next.
- ChatGPT to help translate the framework language into something more relevant to me and it has more freedom to provide some dot connection based on previous chats and context. Using the 3-Question Framework, it helps me identify the core meta-problem close enough we (teams and I) can actually do something about it: “What is really driving these recurring issues?”
- Claude or Gemini as second opinion outlets by describing the overarching problem, 3-question framework. The importance here is double checking the assumptions sections, are they similar or vastly different? What could be the cause of that? What context might they be missing? These are your clues for whether the AI are faking it or not.
- Package findings into two or three slides with symbols, not walls of text.
- Rehearse using AI to simulate pushback so I’m ready when the hard questions come. This condenses weeks of scattered thinking into a few hours. Not because AI does the work for me, but because it keeps me honest and sharp.
What This Is (and Isn’t)
It isn’t:
- Replacing people with machines
- Automating leadership judgment
- Delegating tough calls to a chatbot
It is:
- Augmenting my own judgment
- Clarifying messy insight
- Reducing cognitive load so I can focus on human conversations
Why I’m Sharing This
I didn’t invent this process. I borrowed from systems thinkers, operators, and folks in the AI community. Then I adapt it to my context, where infrastructure is patchy, leadership guidance is not always clear, and where imported frameworks often felt too Western to stick. What worked on one slide deck lost its punch by the next meeting, so I had to reshape ideas constantly to match the rhythms of APAC teams. You do not need to be an AI engineer to be AI fluent. What you need is curiosity, willingness to be wrong, and a processed way of working. Integrate, test, reframe, evolve.
Try It Yourself
This 3-Questions framework isn’t just for checking AI’s answers, you can flip it to yourself, and use AI to equally gain insights about your own perspectives. Next time you face a messy challenge, try the 3 questions with whatever tool you already have:
- What assumptions am I making?
- Why are they true? (Use the Five Why Model)
- What alternatives exist? Notice what gets revealed. These are signals to pursue for clarity, not a sign you’re in the wrong.
Closing Thought
If you walk away with one thing, let it be this: AI isn’t here to hand you polished answers. It’s here to sit across the table, poke at your assumptions, and widen your lens when you need it most. The real value is not about replacing judgment, but about giving you access to perspectives you would not normally have. When you test your thinking across that wider board of advisors, you end up trusting yourself more because you’ve already explored the edges and pressure points before making the call. That is the power of AI for me: not shortcuts, but sharper questions, cleaner definitions, and faster confidence in what matters. And no, it doesn’t translate into efficiency and faster solutions for metrics’ sake; instead you get to arrive together with other humans much further into the challenge and hypothesis than ever before. That is where we start our clearer eyed debates and exploring new solutions. And you might end up surprised meeting those vain efficiency and speed metrics after all, all because of clarity together.