Eric Lee

Staff Software Engineer | Helping developers stay current with tech news and grow their careers through consistency

[Eric Lee @sdusteric]

"The team faced a migration of 33 API endpoints to a new cloud-native architecture"

There is a trend when demonstrating how companies use AI they always use migration as example. We don't migrate stuffs that often right?

https://t.co/4FBE2BjFx0

[Eric Lee @sdusteric]

@grok @TedPillows Is it time to buy AMD

[Eric Lee @sdusteric]

They weren't kidding about token usage during workflow. 14 minutes, 1.4 M tokens

[Eric Lee @sdusteric]

@jarredsumner When can we expect this on public builds

[Eric Lee @sdusteric]

@eastdakota Now we are making content for bots instead of humans lol

[Eric Lee @sdusteric]

@ChadMoran Indeed!

[Eric Lee @sdusteric]

Having lots of tabs open with many agents running doesn't mean you are productive. You are just busy and you are still the bottleneck.

Unless you don't review and just keep clicking accept

[Addy Osmani @addyosmani]

https://t.co/0DyFIXJueI

[Eric Lee @sdusteric]

@dsaccon Good call out. There are couple issues on the confidence level. I can only imagine hitting the balance between token usage and enough context is difficult.

https://t.co/ubaxKnwqsx
https://t.co/r26XoXoZv2

[Eric Lee @sdusteric]

rtk: a Rust CLI proxy that cuts Claude Code token usage by 60–90%.
It filters output from git, tests, lint, kubectl etc. before it hits the LLM context. A pre-bash hook reroutes commands like git status through rtk to strip redundant info. No workflow changes.
https://t.co/RMX6Z7opeq

[Eric Lee @sdusteric]

@optic0n Thanks for sharing. Seems it might get too aggressive from time to time?

[Eric Lee @sdusteric]

@ChadMoran Doing everything not to release it lol

[Eric Lee @sdusteric]

@jjacky You know the answer lol

[Eric Lee @sdusteric]

@Hilbe Yeah it was fun

[Eric Lee @sdusteric]

New Antigravity: Google's version of Codex

[Eric Lee @sdusteric]

If this benchmark is true and reflects actual use case Damn, performs on par or even better than Opus 4.7 at cheaper cost is gonna make people rethink their agent setup I'll switch my agent to use this to test. Also, might share my agent if people are interested even though I haven't shared any details lol

[Eric Lee @sdusteric]

Oxlint without type aware is 78.9x faster and uses 94.4% less memory 🙈
Should still use it with type-aware but this is insane!

[Eric Lee @sdusteric]

@ChadMoran @benmoon @Rivian Who are you thinking to kill 🙈🙈

[Eric Lee @sdusteric]

Always wanted have this range syntax and it's finally in baseline!

https://t.co/I6jOAQ38C3