LLMs learn from the internet.
robots don’t.
someone has to teach them how to pick up a can, rotate it, and put it back.
that’s what @axisrobotics is doing.
you control a robot arm, record the movement, and sign the trajectory onchain.
physical AI won’t be built from prompts alone.
it needs real-world data.
most treasury automation talk stops at "faster settlement"
the more interesting question: what happens when you encode your policy directly into the payment itself and that's what @arc is building toward
two approvers haven't signed? payment doesn't go. balance hasn't hit the threshold? sweep doesn't trigger. wrong time window? payroll waits
not a dashboard setting. the constraint lives in the transaction
what made it click for me: Circle's own treasury team runs global intercompany operations through just two smart contract accounts. not one per region. two. 24/7, no third parties
i keep coming back to that number
still testnet. local tax and reporting obligations don't get encoded away
but reactive treasury chasing spreadsheets, waiting on cut-off times becomes a design choice, not a structural constraint
base just dropped its july recap: 25 ecosystem launches, and the top two aren't consumer apps.
1 morpho shipped fixed-rate lending on base with midnight. borrow and lend at rates you actually agree to, not a curve that moves under you.
2 https://t.co/OY5iHLRUHw launched a b20 token launchpad. in the first 72 hours: 8,045 tokens created, $14m in dex volume, 12k+ holders.
quiet month, loud numbers. this is what "the blockchain for global finance" looks like when it's not just a slogan.
@naimeh70 The Rally comparison is interesting but budget dropping as a gatekeeper doesn't mean it disappears doesn't a new gatekeeper usually show up to replace the old one?
@naimeh70 had a coworker who did this constantly in client calls, making up delivery dates on the spot that the rest of the team then had to quietly scramble to actually hit