Sumit

[Sumit @Desolatte]

Opus 5.5 making changes, Codex ruthlessly poking holes in them. Best multi-model pairing I’ve used so far.

[Sumit @Desolatte]

@jacksunny18 Tried that, but with agents documenting every fact after a while things just bloat up.

[Sumit @Desolatte]

Built a thing for fun: when my app breaks, Claude Code reads the code and tells me why, with evidence.

Runs on your own machine, open source.
https://t.co/oKRnkPl1cj

[Sumit @Desolatte]

@iamvishal16_ios I so started with that.

Having something that tells me this code or behaviour is broken at this particular line is so helpful

[Sumit @Desolatte]

@theo Whats your rating about the code taste

[Sumit @Desolatte]

@theo Not even a better sol model?

[Sumit @Desolatte]

@SwiggyCares And I have been asked for the same details again and again. Have provided all details, escalted the issue multiple times, talked multiple call support executives and still facing delays with no resolution in sight.

[Sumit @Desolatte]

Hi @Swiggy, @SwiggyCares, just wanted to let you guys know the stelar, hard working and efficient support team you have.
I mean only an exceptional team can delay a simple process like FSSAI and Map pin location update for more than a month and still get it wrong.

[Sumit @Desolatte]

@SwiggyCares Latest tickets that I have raised 23248564, 23205242. And from the subject line it looks like they are trying to update pin location for outlet that is already updated.

[Sumit @Desolatte]

@SwiggyCares I have 3 outlets on Swiggy, last month I raised a request to update license and pin location on all 3, but after 4 weeks of delay only 1 outlets pin location was updated, rest 2 are still showing the old location.

[Sumit @Desolatte]

@GithubProjects https://t.co/DmsOw67QhU

[Sumit @Desolatte]

@GithubProjects https://t.co/T0Q3nVctZT

[Sumit @Desolatte]

Maybe they really are scared and all of this not just marketing

[Dario Amodei @DarioAmodei]

We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.

Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.

You can read the full post here: https://t.co/OGyPb7yaYt

[Sumit @Desolatte]

@theo Thinking of switching from my 100$ claude sub to codex. Any hot takes?!

[Sumit @Desolatte]

@theo Here I was using up 50% of my limit with Fable in 2 days. Now with opus i am struggling to reach 70%

[Sumit @Desolatte]

I haven't written a line of code in 2 years.

I still ship more than I did when I wrote all of it myself.

Wrote up what that's actually costing me, what I tried to fix it, and the harder problem I still don't have an answer to.

Link 👇
https://t.co/qy7d80jI8x

[Sumit @Desolatte]

@GithubProjects https://t.co/T0Q3nVctZT

[Sumit @Desolatte]

A reset? 🤷‍♂️ @ClaudeDevs

[Sumit @Desolatte]

@prasenx the one whose agents start unionizing over context window conditions

[Sumit @Desolatte]

@johnennis totally valid. that's why mage-memory only proposes what to remember — a human still approves and commits it. your notes, your git history.