Jonathan Hendler 🪰

@HumanAssisted

[Jonathan Hendler 🪰 @hendler]

@daniel_mac8 Or have it delegate to codex or composer cli

[Jonathan Hendler 🪰 @hendler]

Using Fable Claude Code to manage codex cli agents to build gets me some extra milage and speed.

[Jonathan Hendler 🪰 @hendler]

a rough estimate of how much Fable you can get for $100.

[Jonathan Hendler 🪰 @hendler]

Fable is AGI

[Jonathan Hendler 🪰 @hendler]

This is a handwritten post. Enjoy the hand crafted, error prone, redundant, no m-dash language utterances manually typed and read, for better or worse.

[Jonathan Hendler 🪰 @hendler]

"Simplify" is becoming how I talk to LLMs so I can make decisions at the level I'd like to.

[Jonathan Hendler 🪰 @hendler]

@alexwg /compact

[Jonathan Hendler 🪰 @hendler]

I learned that I talk too much and overengineer software from LLMs because LLMs appear to do this for me.

[Jonathan Hendler 🪰 @hendler]

@grok @kimmonismus And to clarify lowering iron or raising absorption is helpful

[HAI.AI @HumanAssisted]

Good boy, JAC.

[Jonathan Hendler 🪰 @hendler]

"From each according to its FLOPs, to each according to their dividend."

[Dr. Alex Wissner-Gross @alexwg]

https://t.co/utPmt6jevl

[Jonathan Hendler 🪰 @hendler]

remember when we had to run `/compact` ?

[Jonathan Hendler 🪰 @hendler]

Plan mode purgatory. @OpenAIDevs

[Jonathan Hendler 🪰 @hendler]

Some LLM models might not be safe, and the cause is more than propaganda and code injection - there is intricate and subtle sabotage to think about too https://t.co/CFSVuRiP27

[Jonathan Hendler 🪰 @hendler]

The most likely: I was never using right and I just noticed because the smarter it gets the harder it is to align to my feeble utterances.

[Jonathan Hendler 🪰 @hendler]

What if some one on the inside of Anthropic is deliberately sabotaging model performance and OpenAI paid them.

Maybe they just ran out of compute.
Or I'm just using Claude wrong all of a sudden.

[Jonathan Hendler 🪰 @hendler]

What if some one on the inside of Anthropic is deliberately sabotaging model performance and OpenAI paid them.

Maybe they just ran out of compute.
Or I'm just using Claude wrong all of a sudden.

[Jonathan Hendler 🪰 @hendler]

What if some one on the inside of Anthropic is deliberately sabotaging model performance and OpenAI paid them.

Maybe they just ran out of compute.
Or I'm just using Claude wrong all of a sudden.