8 Comments
User's avatar
Sam's avatar

Curious why no mention of Copilot...?

Expand full comment
Caleb Mellas's avatar

I have found copilot to not be as useful. I definitely use it as like β€œnice autocomplete” while coding. But it doesn’t seem to be able to keep enough in context to be as useful for me.

Expand full comment
Jordan Cutler's avatar

Great set of use cases, Caleb. We have so much overlap on how we use AI πŸ˜„

Expand full comment
Caleb Mellas's avatar

That’s fun to hear! If you have any other interesting use-cases I’d love to hear them to keep learning together.

Expand full comment
Stanislav Huseletov's avatar

This is one of the cleanest, realest writeups I’ve seen on practical AI use. Especially liked the debugging + diagram flow β€” hits close to how I’ve been operating.

One point you made really stood out: model sequencing matters. I’ve been running structured ping-pong chains (GPT β†’ Gemini β†’ Claude) across writing, research, and ops docs β€” and it consistently beats single-model loops. Same stack, different order = different outcome.

We tested this across 50+ runs β€” got 98%+ quality in under 20 minutes, vs 2 hours in manual/tab-based setups. If helpful, I wrote up the full workflow here (with metrics):

https://trilogyai.substack.com/p/ai-ping-pong

Would love to hear how you’d extend this to visual/code-heavy tasks.

Expand full comment
Caleb Mellas's avatar

Thanks for your support, glad it hit home!

Interesting, I’ll give your insights a read.

Expand full comment
Kisan Tamang's avatar

I am always curious how people are leveraging AI. so that I can optimize mine. Thanks for sharing this! 😊

Expand full comment
Caleb Mellas's avatar

You are welcome! Glad it was helpful. If you have any other interesting ways you’ve used it, I’d love to hear them.

Expand full comment