Capability and price have scaled together. The inflection comes when “good enough” becomes commoditised.
Users route menial tasks to cheap open source, reserving frontier models for the rare hard problems.
Match Group CEO @spencerrascoff says his company is adding millions in new AI expenses, and many CEOs still aren't sure when the payoff will fully show up.
"So our approach is that we have to pay for these tools. For us, it's $5-10M a year that we're spending on AI tools from basically nothing. So we're slowing down hiring somewhat. That's how we're choosing to fund it."
"In terms of when do we start to see the actual benefit? CEOs, CFOs, and I are starting to ask that question more and more. I think we're benefiting from it, but it's hard to feel it."
From his appearance on the show earlier this month.
Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.
frontier models are accelerating. so is the spend on tokens.
meanwhile prev-gen and open source are getting to "good enough" for a lot of knowledge tasks.
curious what evals and frameworks emerge to draw the line between sota and good-enough inside orgs.
Chat is the right interface for commoditised products (cheap flights). it’s the wrong one for differentiated ones (boutique accommodation). The next decade of consumer ai isn’t chat, it’s “explore” - ai that orchestrates the interface, not ai that is the interface via chat.
"I do not think a chatbot is the right interface for travel or e-commerce." - @bchesky
"I think the future is not apps. The future is agents, but I don't think they're going to be text-forward. I think they're going to be really rich user interfaces."
"Imagine using iMessage to do everything, when in fact every other app has a unique interface."
"With e-commerce, you want a very rich user interface. It would be agentic. You can have a conversation with it, but the point is that it has to be more visual."
Looks like @Wise now charges a flat 61p per FPS amount. Makes it tough to justify as a daily P2P app at the price.
Particularly when most don’t charge on FPS.
@lalitchan@AlexH_Johnson This conflates a few things - lounge discounts typically come out of a fixed yearly cost on a rewards credit card. Lending, if talking a credit card is on monthly interest for balance carried over.
I’m referring just to pure transactional cost at time of purchase.
Whoop getting some serious competition in the no screen wearables space.
Only risk here is if this ends up in @killedbygoogle, the google graveyard, like Google Health in 2012 after only 4 yrs…
JUST IN: Google just launched a screenless Fitbit band to go after Whoop.
Fitbit Air is a $99 wearable built to track your health all day without apps, alerts, or a display.
• tracks heart rate, sleep, blood oxygen, skin temperature, and movement
• no screen, notifications, or smartwatch features
• up to 7 days of battery life
• 5-minute charge gives roughly a full day of use
• works with Google’s Gemini-powered Health Coach
Vibe coding will increase the application pop. size. With that comes both those on the left and those on the far right of the bell curve.
The differentiation to land on the right is taste and articulation.
Latest RFS from @ycombinator: Dynamic Software Interfaces.
Every user gets a bespoke UI, generated on demand.
Flat subscriptions can’t price that. Infra was predictable. Inference isn’t.
Usage-based billing stops being a choice. It’s the only model that survives
Introducing streaming payments.
Track tokens in real-time with Metronome.
Collect payments in real-time with Tempo.
Get paid for every token as it’s burned.
Aligned with @pitdesi here. Contrary to the wider viewpoint that stablecoins are disruptive, they don’t do much when it comes to remittance.
Sure they help offload some ledger management but the bulk of the cost per payment is fx and off ramp for redemption.
How can stablecoins reduce remittance fees to zero?
makes no sense to me whatsoever.
The bulk of remittances $ is migrant workers sending money to family, from US->Mexico, US->India, Gulf->India
Stablecoins can make the money-movement leg much cheaper/near-zero but that isn't the bulk of the cost!
You still need cash-out, FX, compliance/KYC, fraud controls, customer support, and distribution.
When I send money to my cousin in a village in India, there is nothing he can do with USDC. He needs to get the money out into INR and that last mile is where most of the cost lies.
Stablecoins can make remittances much cheaper but nowhere near zero.
the future interface is probably three layers:
1. ambient intent capture
voice, location, calendar, screen context, messages, habits, biometrics, etc. the system understands what you’re trying to do before you explicitly “open” anything or augments your intent deeply.
2. agentic execution
the actual work happens through agents operating software, apis, browsers, documents, email, calendars, workflows, payments, support systems, whatever. most “computer use” becomes machine to machine clerical labor.
3. ephemeral verification ux
humans still need to inspect, compare, approve, edit, reject, or enjoy things. that’s where gui survives but as disposable, task specific surfaces generated for the moment.
When creating a payments offering - the main things to consider when choosing what scheme:
- push or pull payment?
- speed of delivery
- cost of delivery
- customer protection/dispute mechanisms