@jeff_weinstein What will the experience be for the agent owner/user? Will they be prompted to provide the payment when their agent is accessing the site?
@karpathy Now you can use that to train some vertical agents on specific topics! Run a decision model to classify them first and then ground each agent in their category’s knowledge
This curated content will produce some uniquely opinionated agents
My further (informed) fantasies about the near future:
1) No CI. Agents prove each change against rules while they write it.
2) Code is a cache: rules stay, code regenerated with better models.
3) No staging, simulation instead.
4) Humans direct in plain words, agents compile that into formal rules, humans don't read the formulas, only examples and counterexamples.
5) Agents find the rules the human never knew to express (and ask to confirm them).
6) The checker catches contradictions between human's rules while they're still directing.
7) Unchecked/unproved code becomes legacy code.
8) Branches come back as experiments (many versions at once, humans pick among winners).
9) Software gets bred, not built.
10) Slop is variation without selection.
11) If UI is generated on the fly, design becomes the rules every generated screen must follow. A checker runs those rules on every generation.
12) Apps evolve from how they're used.
13) Everyone uses their own software version. One product becomes many personal versions, that stay compatible via shared rules.
@bcherny I literally just checked my subscription page wondering `What the heck is Ultra?!`
This is exciting, but it won't help with how quickly I've been hitting my weekly limit 😅
Good take! I also like that by making the SaaS AI native, you can distribute it in other AI platforms/assistants/etc..
Build MCP/Skills/Plugins, serve them in your own AI agent, or inside any personal AI agent like @Muse@OpenAIDevs Dot or @ClaudeDevs Duet.
That's what Shopify has been doing too with Shop.
The most interesting part about this memo is that at its core, it's advocating for solid operating principles.
This is how any engineering business should operate.
The bottleneck elimination principle is just one example.
I exhausted all my 4 (2 max and 2 pro) subscriptions this week in 6 days.
I’m agent free for the next 24hrs!
Too bad this didn’t line up with the weekend
I liked having one ☝️ agent accessible on the phone, ideally using voice, and let it orchestrate or coordinate with any others I have long running on my laptop/desktop. So I can I have voice calls when I’m on a walk instead of prompting and doing “speed chess” from my phone with all them.