The biggest mistake I made when I first set up my OpenClaw back in February was trying to set up cron jobs to have the agent do a bunch of stuff through LLM. That is a horrible way to go about cron jobs. You want the agent to create scripts and only inject the LLM when needed. Saves you a bunch of usage and everything executes way faster.

Opus 5.5 is AGI. It's literally iterating with me.

I told my claw I wanted it to be proactive and with that instruction plus Opus 5.5, this shit has been amazing

The fact that they're letting them use Claude and they don't even have an internal model for them to use is disgusting

OpenAI has no choice but to swing for the fences. I hope they had insider information about Opus 5.5 and they're going to crush it on Tuesday.

If I had to put it in words, the Codex weekly limit feels like the Anthropic 5-hour limit. Using Astra Except there is no waiting till later for it to refresh. You now are forced to run to X and harass Tibo

Openclaw FTW

Unreal, this Ravens game

Fucking aye man

As soon as my 5-hour opus window resets, I'm going to have to audit everything Astra is doing. Ridiculous

OpenClaw has gotten so much better at running multiple subagents at once. I've got about seven subagents across Astra and Opus all working together, being orchestrated by my main agent, and none of them are failing

James Winston is terrible ....jeez

Very skeptical when I see all of these Grok bot videos and everyone's so excited when the underlying model is so trash

4% percent of my weekly usage is gone already from starting that task lmao

Time to burn