Hot take: AI automation is not magic. It’s iteration.
I don’t know but I usually feel the first version is sucks.
Usually after the N iteration: refining the workflow, teaching the agent what “correct” actually means, and slowly making it useful.
https://t.co/92R7IfNY2a
sometimes i wonder,
if i can delegate 50% of what’s on my plate now…
what would i choose to let go?
and what would i use that extra space for?
more thinking? better output?
or maybe just less mental clutter?
i feel it’s worth thinking about.
do you feel the same sometimes?
AI is only useful when it is attached to a real bottleneck. Not every problem needs AI. But repeated, structured, high-volume work usually needs leverage.
Today we’re launching the OpenAI Deployment Company to help businesses build and deploy AI.
It's majority-owned and controlled by OpenAI. It brings together 19 leading investment firms, consultancies, and system integrators to help organizations deploy frontier AI to production for business impact. https://t.co/GnyjGFaLLA
when code is cheap, it’s hard to justify spending times to think what is worthwhile to build until your product become a gigantic mess (features and code wise)
i think we are all optimising the wrong part which is mindlessly hitting the send button to generate codes.
AI is also so much better at researching, digging through codes, helping human to understand complex systems, writing defensible tests, etc. AI should empower people to do more with human does the thinking AI does the grunt work not the other way around.
i always wanted to bring cafe @cursor_ai to kl and it finally happened last week!
two sessions. 120 people across both. the venue was designed for 50. at peak we had 80 to 90 in the room at once.
we forgot to announce the open mic (where you can just share what you build and connect with other like minded people, like a ice breaking session) 🤣. i was debugging a check-in system i'd built the night before while people were already walking in. chaotic. and genuinely one of the better events i've been part of.
i expected maybe only half of the registrations to show. the venue is far and not very accessible. people came anyway. some just walked in with their friends.
during the event, what surprised me most: @fr4nnyp4ck and @nickwm had been talking to everyone like normal attendees before we announced they were from cursor. most people genuinely did not know. the moment we did, the whole room shifted toward them, and they stayed, talking to everyone, for the rest of the day.
no slides. no keynote. just conversation. the q&a was standing room only and everyone was engaging with the team and asking thoughtful questions.
i've run bigger events where you're technically close to everything but never quite in it. cafe @cursor_ai was different. i actually got to sit with people. hear what they're building, where they're stuck, what they can't figure out. that's the part that matters to me.
talking to the @cursor_ai team, what came through was that they're not watching competitors. they're building what they think is worth building next. composer 2, cloud agent, code review. it had the texture of a founding team still making real bets.
a few things stayed with me:
- a ux professional who came not to build something new, but to get better at how she already works. that's rare at most builder events.
- a claude code plus codex user who saw cursor's model switching for the first time and was impressed with cursor 3 glass agent management view.
- someone who said cursor events always feel warm and welcoming. it's like your neighbor's housewarming. that one landed.
thank you all for coming, wish you had a good experience. i was hoping to talk to more people but was managing logistics more than i planned. see you in the next one I guess :)
ciao
@busyXsanya@dhruvtwt_ Actually in Cursor, you can switch to diff model for thinking and coding. Every model has their own strength, use it and pair it with your goal. Don’t stay in one model.
@bas_fijneman@dhruvtwt_ Tbh, I’m actually quite curious how they hit the limit easily. I’m on $20 plan in Cursor. Code and did product planning everyday, yet never hit the limit.
Yesterday at Cursor Cafe KL exposed something uncomfortable.
Most communities aren’t built for builders.
They’re built for performance.
People pitch wins.
Hide uncertainty.
Optimize for how they look.
But yesterday was different.