Took it to the middle of nowhere and kept signal even when my phone lost it!! 5G with RM551E on a RBM33G. Thermals were good for ~8 hours in 90°F heat!
Unlimited priority TMO 5G for < $13/m...
Need more power!! And aggregation.
Trying to build a ridiculously overkill remote-work system. The requirements:
- battery deployable for 12-24hrs, or run on mains AC
- starlink + 5g + LTE (cat 6-20) WANs with failover and ideally MPTCP for 'bonding' (tbd how that plays nice with video calls...).
- minimal footprint/portable. no permanent cabling. take it from car to campsite. think 1-2 pelican cases.
- must perform in-motion up to ~80mph.
@geerlingguy fun fact: in the ipv6 spec it says that if you use it you are required from then on to be absolutely insufferable to other people who dont care.
Denver's tornado sirens were inadvertently sounded by the fire department on Monday after a 911 caller reported a tornado. There wasn't a tornado. It's the latest in a string of emergency alert errors in Denver. @MarcSallinger reports for @nexton9news:
just remember when the secret service twitter account or whatever posts the next 'bust' its all just a fake outcry.
its harder to buy a flipper zero than one of these. literally just a buncha modems.
@GergelyOrosz There are several subscription products but it's clear they haven't figured it out. Ballpark $20/m? These are published but hard to find.
6mo:
$79 — https://t.co/n7gSpLJfEU
$79 — https://t.co/48LOprJBzF
$89 — https://t.co/y1WiCYTC3v
@MonetSupply@moo9000 It goes without saying that all AI generated code has rigorous human reviews. No one is vibe coding directly to production. We're increasing speed of shipping and innovation, while continuing to raise the bar on security.
@wisplite@usgraphics I think it is the difference between a technical interface made by engineers who were entrenched in the development of the product itself, vs an interface made by designers on a totally different team from the hardware folks.
@rchase@unhurriedself Right, prompt sensitivity (some papers on this out there) can create risks so to derisk I find it's best to hoist stuff out. You can still POC with all-AI and test but then I'd push the agent to eliminate as many variables as you could. Also improves token efficiency at scale.
@unhurriedself@rchase yes exactly.
basically just extract as much as you can to scripts (can do this with agent of choice, mind you) and leverage the LLM where generation is more useful, and consider that anytime the LLM is used, the output could vary by a little or at worst a lot.