kush

building self driving codebases at prev @graphite

[kush @kushxg]

unlimited tokens are a beautiful thing

[pash @pashmerepat]

for the last few months i have been working on long running persistent agents. i believe this is the next paradigm shifting product after chatgpt and codex.

there are a number of challenges, and i'm currently writing a longer post with some learnings & thoughts on why i believe the next big unlock is solving theory of mind

loose thoughts:
- the next paradigm-shifting product after chatgpt and codex is the persistent personal/work agent
- the blocker to making such agents is that our models can't "read the room"
- reading the room sounds soft but it's deep theory of mind
- reading the room is the difference between talking to a bot and talking to an embodied identity you can trust to accomplish tasks and communicate with your colleagues
- reading the room an understanding of who "your person" is, their desires and motivations, and who your audience is
- concretely, it's tracking what i know vs what my user knows vs what the room knows. information asymmetry as a first-class skill
- part of this is understanding compartmentalization. enterprises call this tenting and workspace isolation. "if my user is part of this tent, don't divulge information in a broad channel that's not also tented"
- if i give a friend my home address, i trust them not to announce it in a room of a thousand people. I do not yet trust a model to exercise that same discretion.
- harnesses like openclaw keep trying to solve this at the application layer with permissions, workspace isolation, context injection, and increasingly elaborate agent harnesses. necessary for now, but this likely futile as the final answer. discretion has to live in the model
- side effect: this is also part of why model writing reads as slop. good writing is meeting your audience where they are. same missing skill

tldr: persistent agents live or die on reading the room, and theory of mind may be one of the last major gaps before agi

[kush @kushxg]

who do i need to talk to to get @SlackHQ's message rendering to stop changing

[kush @kushxg]

why would chinese labs ever agree to sign this? deceleration requires collective agreement, which is an unstable configuration given current incentives

[Anthropic @AnthropicAI]

We support this petition, signed by our CEO, several co-founders, and senior staff.

Our own research on recursive self-improvement, published last month, points to the need for tools to deliberately pace the frontier of AI development so society can prepare. We’re glad to see broad agreement across the field. https://t.co/DqwuQfa9xH

[kush @kushxg]

today i got paged because a feature i added... increased usage too much

[kush @kushxg]

@jacobgold 💯

[kush @kushxg]

☕️

[Ben Lang @benln]

We're booking out a local cafe on August 12th in NYC

Grab coffee, merch, Cursor credits, and build with the team

See you there

[lauren @poteto]

one harness to rule them all

[kush @kushxg]

@lottite asleep at the wheel

[kush @kushxg]

turns out saas was never dead! just because you _can_ build something doesn't mean you _should_

[Neha Kalani @thericebowlgirl]

quick update on how this is going: they have gone back to linear because maintaining their internal tool that they vibecoded was taking away from their actual work’s bandwidth.

[kush @kushxg]

@DylanBurnette ah got it, believe we currently only let cursor operate its browser (locally) or its own machine (cloud)

stay tuned for some exciting updates though ;)

[kush @kushxg]

@DylanBurnette cursor already supports computer use! what's your workflow?

[kush @kushxg]

everything at cursor revolves around feedback

- nothing ships until the team loves using it, and they're LOUD when you're not there yet
- how fast can we get ___ into users' hands? how can we do it in half that time?
- where are our product/company/processes failing us and our users? how can we improve, simplify and unship to get there?

[Chris Maconi @chrismaconi]

This is why Cursor / SpaceXAI will win.

[kush @kushxg]

@MerrillLutsky @graphite pleasure doing business with you🕴

until next time!

[kush @kushxg]

feels like we're watching all the avengers assemble for the final battle against thanos

[Jensen Huang @JensenHuang]

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.

AI will transform every industry, power every company, and be built by every country.

Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.

The world needs both frontier closed models and frontier open models.

https://t.co/AUKzoQ5Ikb

[kush @kushxg]

and we're still cooking 😉

[˗ˏˋ Jesse Hanley ˎˊ˗ @jessethanley]

Bugbot.
Automation.
Cloud agents.
Cursor mobile.
Slack integration.

@cursor_ai has eaten the entire development pipeline for me. It’s replaced many other side apps I was using (Greptile, etc). It has completely changed how I work and I’m so grateful for it. The economics now make more sense with Grok 4.5 as I don’t really consume much on-demand credits anymore.

@opencode/@thdxr crew seem to be focused on a different segment that isn’t me (basic dev). @opencode was great when I was stuck at my computer but so much dev for me is now done in the cloud or via mobile. Theres no like 1-click @opencode cloud I can setup. There’s no easy mobile setup. I don’t want to manage another server.

I still use it for quick stuff time to time but yeah Cursor fully won me back due to new workflows.

[˗ˏˋ Jesse Hanley ˎˊ˗ @jessethanley]

Bugbot.
Automation.
Cloud agents.
Cursor mobile.
Slack integration.

@cursor_ai has eaten the entire development pipeline for me. It’s replaced many other side apps I was using (Greptile, etc). It has completely changed how I work and I’m so grateful for it. The economics now make more sense with Grok 4.5 as I don’t really consume much on-demand credits anymore.

@opencode/@thdxr crew seem to be focused on a different segment that isn’t me (basic dev). @opencode was great when I was stuck at my computer but so much dev for me is now done in the cloud or via mobile. Theres no like 1-click @opencode cloud I can setup. There’s no easy mobile setup. I don’t want to manage another server.

I still use it for quick stuff time to time but yeah Cursor fully won me back due to new workflows.

[kush @kushxg]

long compute

[Pavel Asparouhov @Pavel_Asparagus]

You can predict someone's opinion on the chinese models with 100% accuracy by comparing the size of their bags in the labs vs the size of their bags in applied AI cos

[kush @kushxg]

a better alternative to "fable for everything"

[Cursor @cursor_ai]

Introducing Cursor Router, our intelligent model router that selects the right model for the task at hand.

Router delivers frontier-quality results at 60% lower cost.

[kush @kushxg]

@ayushxpatne all 70 of them on the big screen

[kush @kushxg]

maybe it's cope but i think the extra millimeters were worth it

[kush @kushxg]

if you really can't wait... https://t.co/YJl2nSOx6a

[Cassidy @cassidoo]

I can't WAIT to get stacked PRs into everyone's hands! So close y'all.