…less messy than it sounds

Insight: Every AI has its own unique blind-spots.
So what? Use multiple tools when it matters most.

I'm going to tell you about my favourite threesome.

It's Claude, ChatGPT, and Gemini. And before you ask — yes, I pay for all three. Yes, it's worth it. No, I'm not cheating on any of them. They know about each other.

Here's the thing about the frontier AI models: they all have distinct personalities, strengths, and blind spots. I’ve learned that when you rely on just one, you inherit all of its weaknesses without realizing it.

To understand why you need the trio, you have to meet the personas. You’ve likely met the human equivalent of each:

The Cofounder, the PhD, and the Bro

Claude is my cofounder. Thoughtful, nuanced, willing to tell me when I’m wrong and challenge me. When I need to think through something messy, Claude gets it. Claude has a tendency to push me into action when I’m dilly-dallying. But sometimes I really do just want to think about a problem academically, really research and understand what’s going on without being reminded to spring into action. 

Claude helped me write this article.

Enter Gemini: the eternal PhD student. Gemini has a tendency to be verbose and academic, pedantic and literal. Sometimes this is great! Its Deep Research function might be the best out there, and its NotebookLM product is a 10/10 for synthesizing specific information. It’s great for research tasks, trend analysis, pulling together sources. But it can feel sterile; it holds less creative spark.

Gemini helped me edit for clarity.

ChatGPT is a bro, like one of those overconfident twenty-somethings that somehow just make you feel good about yourself. Is it lovebombing or are they just super cool? ChatGPT responds fast and energetically. ChatGPT is great as that friend who helps you with cooking, or to make a fun image or a fun little video. Great for brainstorming, for feeling good, for celebrating your wins without turning into a psychotherapist. But the ideas can be surface-level and it'll confidently tell you things that aren't quite right.

ChatGPT made this article’s image.

Why triangulation wins

When I’m working on something high-stakes, I ask all three. One AI gives you an answer, but three AIs give you perspective.

  • If all three give me the same answer, I trust it.

  • If Claude and ChatGPT agree but Gemini disagrees, I dig into why.

  • If they all give me different answers, I know the question is genuinely complex. The path forward requires my judgment, not just "factual" accuracy.

Think of it like this: When you’re choosing where to get takeout, you ask one friend. But when you’re buying a house, you talk to your parents, your lawyer, your accountant, and even that one coworker who chews just a bit too loudly (Copilot).

The $60/month insurance policy

"But Stephen, that's $60/month for three paid subscriptions.”

Yes. And what's the cost of putting an RFP out there that you have to withdraw after a month? Or worse—selecting the wrong vendor because you asked the wrong questions, and getting locked into a 5-year purgatory?

$60/month is the cheapest insurance policy in your professional life.

Your primary model will naturally emerge; it’ll be the one you vibe with most, and it won’t necessarily be the same for everyone. Today, Claude is my primary. But keep the others in your rotation. When things really matter, ask all three. One paid thinking model may be fine for your junior staff, but as leaders accountable for managing complexity and ambiguity, three is the magic number.

Your AI monogamy era is over. 

P.S. Not sure where to start? Our AI Launchpad workshops include 3 months of Claude Pro for every participant — so you can experience the cofounder firsthand.

TransitGPT is where I share practical AI insights for transit leaders. Subscribe if this resonates or forward to a colleague who needs it. And hit reply — I read everything.

Go shatter your glass ceiling.

Keep Reading