That's a lot of work for that result when a simple text to video would've sufficed. Good AI video is still hard. Certainly not as hard as with an actual crew.
@elonmusk You had mentioned this months ago. Can we get an update? If you can go to Mars, you can certainly do this. You'd make millions of people very Happy! Right @Jason?
seeing a lot of people are grok @bot-curious.
DM me if you want to swap your other sub(s) for a free month of Pro+!
will try to support as many as possible over next 24 hours.
That's exactly how I am using my Hermes agent. I have one orchestrator bot. I dump everything to him and he decides which sub agent works on it. The others do the work. I can keep talking to him. Have others do different work. He reports to me when they are done. Shared GBrain.
between this and @grok bot it feels like we’re going in a direction i don’t really understand. feels like we’re overcomplicating things for most users
- what i want is one chief-of-staff agent. i only see and interact with one thread and dump everything into it. it then routes the work to the right “subagent” profiles, lets them communicate as needed, and brings updates back to me
- each subagent can have its own memory/model like these bots (don’t think skills need to be separated by profile though?)
- but the key is i don’t want to think about which agent to use. once set up, profiles should be mostly hidden. they should feel more like tools/plugins you configure once and forget about
- ideally profiles can also just be created on demand as part of a task since they’re really combinations of existing primitives anyway
i think multiple, composable agents are the right direction, but i’m much more bullish on a future where ~all communication with them happens through a single master thread.
closest examples today are probably @jxnlco’s codex chief-of-staff setup (https://t.co/rLWycyR5QH) or @ampcode’s Puck
@ageshah@iruletheworldmo@mntruell My codex, claude code and my Hermes all share a GBrain and use obsidian with qmd so they all know what they other does. They can use cli and talk to each other also. @garrytan
My kids still cringe whenever I try to use chat in the car or at home. What can we do to get them engaged and get them excited about A.I.? They think it's creepy.
cool use case of chatgpt work i heard last night:
connect your family calendars and explain your kids' interests.
every morning for the drive to school, have it make a podcast that talks about one kid's soccer game that afternoon, one kid's upcoming birthday, some news, etc.
Grateful to cap off the season with my Take Over Basketball teammates at the Nike Tournament of Champions in Chicago! I finished with 84 points in 6 games (14 PPG). Thank you to my coaches, teammates, and family for all the support. Back to work! 🏀💪 @niketoc@takeover2029
Amazing! I still think that the marriage between the audio and video in most of these models is still not there yet. It is still in the uncanny valley stage. It's a real telltale sign that something is AI. Our brains just know that that voice doesn't belong to that person.
Facial expressions, eye gaze, voice, and body movements are synchronized with the spoken output.
The system maintains character consistency and facial responsiveness across the entire session.
What’s the best open source version of Claude Cowork that a) allows you to plug in @OpenRouter or @bittensor b) works with local models and c) is optimized for the average knowledge worker (Hermes and OpenClaw too complex for normies)
Right now Perplexity computer is my best choice (doesn’t do local models yet)