What are you most excited / looking forward to using CoreTex for? Is there an immediate use case to try with it, that you think its uniquely capable for?
As for immediate use cases - I think the most powerful thing right now is just the ability to play with the config with system and agents. You can easily compose with them, give it a prompt, no lock in to any single provider. And the review gates are nice. Code gen is still experimental and will be finicky.
I think what CoreTex is uniquely poised for is being the most lightweight and easily extended "control center" of various AI systems. That's why I doubled down on UNIX philosophy, inherent security, and minimal dependencies for maximal portability. I want this thing to work on any system, take a bullet, and still stay up to help you control AI + complete your goals.
Thanks for your question!
Real brains don't run on respond cycles. Do you see this getting a heartbeat system or is there anything implemented / planned to make it a persistent daemon?
like it could check different surfaces for deviation for useful discoveries, or problems that need to be remediated, or even browse Hackernews for exciting new tech to learn about and grow!
One of the more interesting features I want to implement is the "distributed spine." I want CoreTex to live on multiple machines, offloading particular parts of computation to say - the gaming computer with the GPU running higher local models, sensory data being routed from embedded systems, and a low-power alternative for it to sync to. When the gaming computer turns off, the low power unit will promote itself to primary CoreTex for overnight watching / sleeping / dreaming, then revoke control back to the primary computer when it turns back on. In this sense, CoreTex will be a distributed system, increasing its stability and uptime.
You should be able to use this for anything, really, given its UNIX philosophy and OSS.