A SNOW/DBX acquisition could still happen, of course. But I really hope it doesn’t, and that this gives the dbt/Fivetran combo (Sixtran? dbtran?) the gravity it needs to stand on its own long term.

they just need to acquire a compute platform…

cool, done. if you send me your personal address i can have it forward to you.

i can make an arynn@delphihq.com alias if you want lol

i want them to put max in a racing bull and see what happens

though i’ve been saying this for like a year and a half and clearly been wrong so far. but the gpt-5 launch feels different.

there are non-bearish readings of gpt-5 that may very well be right. but this is the first time the vibes have really shifted.

lol, mostly because “forward-deployed” implies working with customers, vs “solutions” which doesn’t really mean anything at all

fwiw i started out in a role like this (we called it “Product Operations” at LiveRamp) and it was a great place to start. But we mostly were semi-technical, not CS majors.

isn’t that just a fancy name for what used to be known as “solutions engineers” or “solutions architects” then?

imo FDE is more descriptive, but i don’t think it’s a new thing in terms of the work being done

why would you go through the frontend for MCP?

i like to just put the leaves in there and pour through a strainer if needed

> You may not provide the software to third parties as a hosted or managed service

The elastic license is pretty straightforward; i think dbt would be able to enforce it against the Dagsters of the world if they want to.

I assume they’ll come to some kind of agreement with them & others.

not sure about the experimental design here. it seems like the treatment group spent 3 more hours per week learning english and the control group just got… nothing?

i would have liked to see the control be a (non-AI) group tutoring program or something. but still.

with openai (and maybe some other model providers) you can use seeds to get mostly deterministic responses