the labs need facebook's '2g tuesdays' but instead of slow internet, one day a week employees get normie token limits

👀 we have time to visit a few more companies today! dm if you're a small sf startup and want to give your thoughts on an early demo! we'll come to you :)

if you’re a designer at DevDay @vinvan and I would love to meet you!

who’s at OpenAI DevDay on Tuesday? i’d love to meet!

.@jsngr and i are in sf this week for devday! we’d love to stop by small startups and show you something :) dm if interested!

Sharing my first of hopefully many research blog posts! jasperlu.com/blog/training-s… This one is the kind of educational blog post I wish I'd had when I started with RL for LLMs. I tried to make it as open as possible. Every rollout is browsable, the code is open source, and I walk through my entire thought process, from learning rate sweeps to reward shaping.

any boston restaurant recs for a date night?

Excited to share @profound has raised a $180M Series D, co-led by @sequoia and @kleinerperkins. Winning in AI Search requires an inhuman amount of work. That is why we are building the AI platform for marketers, underpinned by two things: 1/ Your AI Marketer, a marketing expert that proactively investigates your data, identifies opportunities, and then does the work while you stay in the loop. 2/ Context Manager, a living synthesis of your meetings, email threads, and brand data powering this new teammate. As your brand evolves, so does your AI Marketer. AI labs are pushing the frontier in fields like engineering, law, and finance. But marketing, one of the economy's most complex and consequential applications, has received far less attention. That’s why we’re expanding our applied AI research team. We’re bringing together research scientists, engineers, and IOI medalists to advance AI for marketing. Their focus: post-training models for marketing workflows, and building the world's first AI benchmark for real-world tasks in the field.

i don’t love how development is now pay-to-win. life is easy as long as you can afford fable and can arena everything. if you don’t, you’ll waste time thinking about limits and getting cheap tokens. some days it feels like playing a videogame with insane microtransactions.

a systems reading group but instead of a paper we discuss a Bret Victor video every week

there's no sadder feeling than finding out your agent stalled overnight

given today's news, a great friday night watch is this 1996 docu on fermat's last theorem: dailymotion.com/video/x223gx… my math teacher always told us to think of Andrew working on his proof for seven years (!) whenever we complained about working hard. to this day, i still do :)

i love grokbot but i want to spend 0 time thinking about what bots i should have: whether it's better to have a "chief of staff" bot or three bots that together fulfill the role etc. poteto poteto poteto, can you just (with my consent) mine my gmail/connected apps and figure out what grok bots i need? plus give me a claude/chatgpt prompt i can run for you — if you need to learn about me beyond mcps. bonus points if there's always an orchestrator grok bot by default. the idea that i have to pick the right bot to talk to gets tiring!

let’s see what happens

if you buy the idea that we’ll stop looking at code… it follows that the ui of the next github shouldn’t be a list of files and folders

introducing pcstack (lmao), a way to use pstack in conductor cloud: gist.github.com/vincentmvdm/… i'm a pstack disciple but also love conductor cloud. this skill teaches my agents how to map pstack's cursor concepts to conductor. install it and make sure pstack is in skills/vendor

trying something tonight to save $: hooking up the conductor cloud mcp to cursor! my orchestrator has to be a cursor cloud agent since i want zero risk of it dying overnight due to usage limits. but i'll save a ton of money if the agents it spins up use my subs via conductor :) i didn't want to sacrifice quality though, so i also made a pstack 'adapter' for conductor. it teaches agents how to use pstack outside of cursor. so e.g. /poteto-mode still works. besides cost savings, the nice thing is that my cursor cloud agent can route to my codex or ant subscription based on usage. and that even if my subs run out, the cursor agent can technically complete the job itself. not sure if this will work but it'd save thousands of dollars for long runs!

ai notes 08/20: - pstack is the best thing i've tried in a while - giving grok bot linux versions of apps is underrated - mine dictation transcripts for factory improvements - prompting while z2 running is heaven