Prompt engineering for Reactor is so freaking satisfying.

Found a method that boosts speed by 2-3x while also improving quality of the AI ghostwriter even further.

Fewer instructions. Fewer steps. Better outputs.

Glorious.

If you’ll be my BodyguardGPT 🎶🎵🎶

Hmmm…. what guilty pleasure content should I like first….

Okay here's a fun one.

Just optimized a prompt that consistently does better on Claude Haiku than Claude Sonnet even.

There's so much about these LLMs we have yet to understand....

The AGI emergence has begun...

We all have those things we want to hide.

That weird gap in your resume.

Those financial struggles you’re too embarrassed to admit.

Those extra pounds from not being in your 20s anymore.

For me? It was my hair.

Basically every day for 3+ years I’ve worn a hat.

Outdoors.

Charlie just killed the viral AI post template

I've been using ChatGPT for tips related to my new hobby lately.

Fascinating what it decides to store in memory.

Relevance > Cleverness

Looks legit. Guess I still got it

Dog moms at my morning park are using ChatGPT every day at work.

The 3 PROBLEMS regular people asked about

(and the prompt FIXES I gave them):

First, the video has my cute dog Olive (8 months old doberman mix) playing with a little 4-month-old pitbull named Finn.

Sorry I

Thinking about making a batch of tshirts with this on the back…

Say what you want about the rebrand, but Nathan's still got it.

Taking the time to reply to a little post from a customer.

No doubt this culture is why (convert)-Kit is winning in such a crowded space.

I wonder if he know's I spend $2000+/year with him

I've gotten superior results from Claude Sonnet vs Opus off of a huge megaprompt in three, totally different cases/prompts.

Can't wait to figure out why the smaller, less-capable model could possibly think and write better.

My Reactor co-founder Shawn's theory:

'Over-thinking'