I want to share a few things that have been happening recently that all point to the same underlying shift — how AI is moving from being a tool you use to something you actually collaborate with.

AI as a Copilot (Not a Replacement)

First, a personal example.

My wife recently started a yoga account on Instagram and TikTok called @tracingthepose, and it’s been really interesting to watch how she’s using AI in the process.

Not to replace her teaching — that part is still very human, embodied, and experiential — but to help her structure, plan, and sequence her lessons.

She uses AI to think through:

• how a class should flow
• how one pose logically transitions to the next
• how to explain concepts clearly to different levels of students

What’s fascinating is that it’s not diluting the practice — it’s sharpening it.

AI isn’t telling her what to teach.

It’s helping her think more intentionally about how to teach it.

That’s the pattern I keep seeing: AI supporting craft instead of replacing it.

(If you’re on Instagram or TikTok, definitely give @tracingthepose a follow — it’s a great real-world example of this in action.)

Orchestrating Intelligence, Not Outsourcing It

That ties directly into the second thing I’ve been spending time on.

I’m currently working on a few AI projects, one of which is inspired by an idea first posited by Andrej Karpathy — the concept of an LLM council.

Instead of relying on a single model or a single “answer,” you orchestrate multiple models or perspectives together — almost like a board of advisors — to reason, challenge assumptions, and synthesize better outcomes.

What I love about this idea is that it mirrors how good human decision-making actually works.

The goal isn’t speed at all costs.

The goal is better judgment, better context, and better synthesis.

Again: AI as a collaborator — not an oracle.

Asynchronous Conversation, by Design

Which brings me to the third thing.

The latest episode of the asynchronous podcast is underway, and this time it’s with Joe Polish.

If you don’t know Joe, he’s a longtime entrepreneur, marketer, and connector — founder of Genius Network — and someone who has spent decades thinking deeply about relationships, leverage, collaboration, and value creation.

What makes this conversation different isn’t just the guest — it’s the format.

This isn’t a one-hour, live, polished interview.

It’s an ongoing asynchronous conversation, unfolding over time.

That means you’re not coming in at the end.

You’re getting in at the beginning — while ideas are still forming, evolving, and branching.

And because it’s async, the conversation stays open.

If you hear something that sparks a question or a direction worth exploring, you can send me a private Carbon voice message and influence where the conversation goes next.

You’re not just consuming it.

You’re participating in it.

👉 Listen here as the conversation unfolds:

https://carbonvoice.app/c/693afb281f0103196ed224ae

The Bigger Pattern

When you zoom out, all of this points to the same theme:

AI is most powerful when it augments thinking, structure, and judgment — whether that’s teaching yoga, designing systems, or having better conversations.

And asynchronous formats — in work, learning, and dialogue — create the space for that kind of thinking to actually happen.

CTA (Killer, but On-Brand)

If any of this resonates — AI as a copilot, async as a force multiplier, or designing better conversations — send me a private Carbon voice message.

I’d love to hear what you’re experimenting with, what’s working, and where you’re getting stuck.

Make it an effective day, everybody!

Ari