Be careful out there folks. Someone created a Remitly account under Alyssa's email, name, & address, and attempted to send $1,300 to someone in Mexico from a random Huntington bank account. Wild.

Alyssa put a preorder deposit on an R2 over two years ago. Today, we finally got to test drive one… and place our order! Super excited. Seems like it’ll be a great SUV (and first EV!) for us.

Playing around with Jev, and it is now classifying my email... for about $0.0001 per email thread. About 6x cheaper than Luna, 60x cheaper than Sonnet 5/Terra, and 150x cheaper than Opus/Sol doing similar classification. Will need to gauge quality of the classification, of course, but the utility of this type of model is pretty clear. Cool stuff.

You get on a role with AI, ripping through features and fixes, only to be brought back down to earth by it fumbling an obvious implementation that should use an enum. 🤦

Darn it Apple. I think I might want an iPhone Duo. 😬

The Shelton/Alcarez match at the US Open was insane. Sooo good. 🍿🎾

Just a few agents on this session... 😅

This is getting pretty cool... my latest meeting notes were auto imported, Boz assessed the AI summary and validated/fixed facts, and automatically updated other active sessions/docs based on the conclusions from the meeting (all in a way I can observe & approve/veto).

My idea is starting to pull some weight. This is one agent needing a decision to be made (aka, pushing the "AskUserQuestion" sdk interrupt that would normally pause the h-i-t-l session), while a separate agent used my past actions, preferences, and decisions to make an informed, observable judgement & decision on my behalf, all without me lifting a finger.

What started as a silly "I wonder if..." exploration project has turned into some serious cooking. For now, a teaser...

Got to properly hang out with my Laracon bestie this year. @sm0olr, such a cool guy. Unfortunately, no longer a Laravel dev. It was fun while it lasted, dude! 💔✌️

After 4.5 years of living 500 miles from the nearest family member, Alyssa and I finally got to experience the “grandparents are watching the kids on a random Tuesday night so we could play tennis” babysitting strat. It was awesome. Go grandparents. Y’all are the GOATs. 🙏🙌

This is an awesome shipping option. I love it when small businesses lean into their personal identity.

Another fantastic Laracon! Great hanging out with friends and making new ones. Can’t wait till next year (or maybe an international Laracon 🤭)

Super early thought, haven't fully worked it out yet, but: I'm feeling bearish on closed model providers right now. I just feel like the performance gap between open & closed is close enough that the perks of 1) control over access & hosting (fewer rug pulls from outages or others' decisions), 2) security/privacy, 3) cost savings, and 4) the ability to fine-tune the model into a specific way of doing things (making it have better output *for a specific org*, likely at better efficiency) will outweigh the cons of locking your org into a closed provider. Like, if Walmart can deploy a better custom model *for their specific needs* than a "tuned for everything" Fable 7 in the coming years, with lower hosting costs and complete control over it, why would they choose to use Anthropic at that point? Or, if I could use a tuned K3 today that was super good & efficient at the tech stack/tools I use by not wasting parameters on current celebrity gossip, and every interaction I had with it (or my dev community at large) made its core inference better over time for me specifically, I think I'd give up (or reduce) my Claude subscription for this. I suppose this is what Cursor is doing with their model, so I'm stating the obvious here. I guess this explains the hardware race. The models will be a commodity, the hosting will be the margins. Same story, different day. The models right now are just a capital raising play for hardware investment. Hmm. </thought>

Seeing what people choose to wear for airline flights is fascinating.