Leo Kwan

software engineer at snap | prev at instacart, square | building voice ai tools on the side

[Leo Kwan @leojkwan]

@thsottiaux hero

[Leo Kwan @leojkwan]

5 hour limits are awful especially as you orchestrate with frontier models by the time the goal picks up and lanes get in their groove. The limit is hit. And when the limit lifts u have to pay the cost of catchup.

only benefit of the 5 hr window is sometimes u guys let the goal run for longer pasge 0%, so more windows ~> more of this token arbitrage . But I think u guys patched this recently

[Leo Kwan @leojkwan]

@OpenAIDevs big market grab on whichever company does /goal better, very excited for this

[Leo Kwan @leojkwan]

@trq212 could not agree more , a GitHub pages & Jekyll resurgence

[Thariq @trq212]

A few end of week ships:

You can now set effort to 'max' which reasons for longer and uses as many tokens as needed. This will spend your usage limits more quickly so you have to activate it per session.

Hit /effort to try it.

[Leo Kwan @leojkwan]

Claude spawned a dream team of Agents based on my skills to code review and run e2e tests on our Shopify site. we lost ultrathink but the highlighted texts return! @trynia_ai

[Leo Kwan @leojkwan]

new codex app needs to support existing skills /slash commands from ~/.cursor / ~/.claude directories if they want better adoption

[Ryan Dahl @rough__sea]

This has been said a thousand times before, but allow me to add my own voice: the era of humans writing code is over. Disturbing for those of us who identify as SWEs, but no less true. That's not to say SWEs don't have work to do, but writing syntax directly is not it.

[Sam Altman @sama]

New ChatGPT memory improvements are awesome!

[Leo Kwan @leojkwan]

"Ultrathink no longer does anything. Thinking budget is now max by default."

why do I feel like we lost something special still

[Leo Kwan @leojkwan]

@flexibits iOS app is completely bricked in latest version

[Leo Kwan @leojkwan]

@sama Comet is that you

[Leo Kwan @leojkwan]

sonnet 4.5 crazy good

[Leo Kwan @leojkwan]

Try Nia is such a huge productivity boost to Cursor. Latest docs from external toolings (https://t.co/f2kJdu4dxo, supabase, claude) is indexed. Best part is setting everything up within Cursor Agent itself. impressive job @nozomioai

[Leo Kwan @leojkwan]

@joaoaguiam which model's working better for you?

[Leo Kwan @leojkwan]

best AI + tooling by use cases:
opus 4.1 + cursor for dev
comet + its assstant for web browsing
sesame voice for todo list/venting while driving

need to test outgpt-5 high + max before it rolls over to usage based pricing on cursor

[Leo Kwan @leojkwan]

gave cursor another gamble since opus 4.1 came out today. after it generated this interface 💩 (with an existing roadmap & clear todo docs).... I'll stick with claude code for now (but on cursor still).