building infrastructure for things to happen IRL with @itmstudio tweets are my own
ꪀꪗᥴ
Joined September 2012
- Tweets5K
- Following2.7K
- Followers2.4K
- Likes18.9K
one thing i have been thinking a lot about is the moat of a “bank” and what an effective neobank looks like. no bank today runs the work and running the work requires building around a customer archetype. why aren’t more banks running the work behind an industry with all the new capabilities we have around AI?
What Jack meant by there is no moat is that many of the future companies will eventually look very similar even though they all started in completely different sectors and you’re already starting to see bits of this now
Square e.g Payments -> Lending & Payroll -> Business Operations -> Applied AI Lab
Ramp e.g Corporate Cards -> Expense Management -> Back Office Operations -> Applied AI Lab
Stripe e.g Payments -> Financial Infrastructure -> Applied AI Lab
Shopify e.g Online Storefronts -> Payments & Fulfillment -> Merchant Operations -> Applied AI Lab
What’s interesting is that all these companies started from completely different entry points, but over time they’re starting to look more and more similar and eventually will compete with each other in some form
I think ai is going to accelerate this, making it much easier for companies to expand into markets that historically would’ve taken years to enter
My take is that companies are going to look more like pods of interconnected businesses vs operating in one singular sector and I believe a lot of sectors will start to merge
e.g there won’t be a fintech sector, it’ll be a merge of finance, commerce, software and intelligence all bundled together. The same could happen with compute, energy, and finance, or media, commerce, and advertising etc
And as these sectors merge, companies that never competed with each other will start competing for the same customers
Think companies that are extremely ai forward are thinking about how many sectors they can bring together under one company vs the ones that have been skeptical or too slow to move
the better the models have gotten the more i find myself chatting in the worktrees, spending 90% of my time in plan mode, really ironing out the full testing matrix, etc.. before letting it write a single line of code. in some ways, i find after you let the LLM start writing code, it becomes unreviewable. even with code review and whatnot – its just too easy to give up and merge it in.
coop has crushed it in japan... didn't really know what it would take to support japan for the first time when he reached out but very cool to see how his project keeps growing and the lineups keep getting bigger and better
gotta hand it to @zackfox one of our earliest users of the platform for selling out knockdown center one of the largest venues in new york by building his own audience on @itmstudio across performances, popups, comedy nights, movie premiers, etc all over the us, then leveraging the data to do everything from direct comms to building hyper specific lookalike audiences on platform
so many people reached out in my dms. i think this experiment is worth it as it hard proof that you can start a new type of ai native smb. anyone looking to get involved, particularly on the upfront capital side, pls reach out. we have an expert network of operators we can tap.
is anyone here interested in opening a new music venue in nyc? i think with everything we've built for experiential business operators + ai i feel confident we can run a venue, which is traditionally very ops and admin heavy, at a much better margin. but it has to be an AI native project from the ground up.
since llms have gotten better i’ve been far more expressive in my use of enums rather than raw text strings to represent data. it’s become infinitely easier to type a wide range of possibilities for values, and that has opened up a far better ux.
4000+ attendees in paris this week for TONES OF PARIS by @colorsxstudios.
COLORS has 8.6mm subscribers on YouTube and a verified IRL audience of 25,000+ people across New York, London, Paris, Mexico City and Madrid as they use @itmstudio to go experiential. and this is all in under 6 months.
it should probably be in the west village
is anyone here interested in opening a new music venue in nyc? i think with everything we've built for experiential business operators + ai i feel confident we can run a venue, which is traditionally very ops and admin heavy, at a much better margin. but it has to be an AI native project from the ground up.
is anyone here interested in opening a new music venue in nyc? i think with everything we've built for experiential business operators + ai i feel confident we can run a venue, which is traditionally very ops and admin heavy, at a much better margin. but it has to be an AI native project from the ground up.
sold my bitcoin after yesterday’s hack … i don’t see how it doesn’t get hacked given how fast we went from “count the number of r’s in strawberry” to solving navier stokes
so what are the best alternatives to storing env vars securely? something like infiscal wouldnt help here. i guess its really that your cloud provider should spend the most amount of time and resources making sure your env vars dont leak. which is the absolute doomsday scenario.
i don’t think people on here realize how much of the world has moved off trad employment here. so many people are casually running experiential businesses now. classes, workshops, lessons, etc etc
my personal thesis on work agents is that it only makes sense for executive functions (at least how they operate today on chat + harness). “work agents” for the people probably just look like 100x smarter outcome driven saas tools that they are already used to working with. and that requires a robust ontology that won’t get absorbed into a frontier model any time soon.