Guess I’m setting up a bunch of Hermes profiles for my spare MacBook and going through OpenRouter w/ Muse Spark and making my own “Muse Bot” without the killer UX behind Muse. But we can just connect them all to slack and have Muse orchestrate them anyways. So many dang options.

Most of you have probably already done this, but try creating a dashboard and MCP server so your agent army can communicate and collaborate. For Muse, simply provide your login credentials, no connector needed. It’s def a time saver compounding and keeps them all kind of on scope

Cool little tip -> Muse control your slack, talk to your Hermes and collaborate and handoff files through Google WS. Just stack my ideas and thoughts into notes app throughout the day, @Muse delivers jobs to the 3 agents. Fun stuff.

Makes me want to make a muse pet app for the Meta Displays.

POV: riding my e-bike, listening to Joe Rogan talk about aliens, and playing brick breaker in my sunglasses at the same time. We live in an insane timeline and I'm not even mad about it. Built this myself for the Ray-Ban Meta Displays. Pong, brick breaker, blackjack, some puzzle games, the works. It's free if you have the glasses, just add arcade.bishoptech.dev as a custom site in the Meta AI app. It’s built and hosted on @vercel. Not only can I control my Muse from my glasses, but my little Hermes too. Nerds rule girls drool @Meta will take over wearables for sure.

Bike rides have been so much fun lately, when I get some more people on here locally and get to play with the ‘capture the block’ feature it’s going to be even more fun. Get your kiddos building apps, it increases their imagination in the world of tech I think. Takes more to impress kids nowadays I feel. The smart ones anyway 😎

I love that we’re getting all these new flagship models, but I’ve somehow gotten to the point where I’ve orchestrated this little army of Luna agents that can handle almost anything I need for work 😂 Some of the jobs I give them literally span 24+ hours, and somehow it still barely puts a dent in my usage. The output has honestly been surprisingly close to what I get from much larger models too. Granted, with design stuff I might have to iterate a couple times to get exactly where I want, but I eventually get there. It’s also made me wonder if super cheap, super fast cloud models might actually curb some of the push toward locally hosted models. If I can have cloud agents running insanely fast, getting way higher tokens/sec than I could reasonably run locally, putting zero strain on my own hardware, and seemingly running almost nonstop even with a basic Business seat… at what point does building a dedicated local AI server stop making sense unless you have a really specific reason for it?

Seen this remixed several times with dogs but when you put it on a baby it’s 10x funnier

Can you guess which month I turned in the time card for the hobby?

*adds 6 months to Apple app review process for design bugs*

Just got absolutely wrecked at Moonshine beach in Branson MO. Just got done with nice little float away from the screen and terminal and THIS IS WHY I STAY INSIDE AND CODE ALL DAY