Idrees Kamal

AI Engineer by day, custom tailor by night

[Ali Spittel @ASpittel]

i really don't want to use your agent, i want to use my agent to use your thing.

[Idrees Kamal @signedbykamal]

A well crafted AGENTS.md is truly all you need

[Token Gremlin @TokenGremlin]

Best AI subscriptions right now:

1. ChatGPT
2. Grok
3. Cursor
4. Literally save your money

If you’re paying Anthropic, cancel it.

[Idrees Kamal @signedbykamal]

@bot literally just needs a voice orchestrator and it might just replace most of my codex usage

[Idrees Kamal @signedbykamal]

@avenger8474 @klinkicz @ForwardEditor "better" output depends on the task. If it's complex and ambiguous, stick to Sol Medium or High. Otherwise, Luna Max on Fast mode wins on cost and speed

[Idrees Kamal @signedbykamal]

@AdamHoltererer @thsottiaux A loss leader doesn’t mean you’re losing money…. 🤦‍♂️

[Idrees Kamal @signedbykamal]

@AdamHoltererer @thsottiaux Luna is positioned as a loss leader for OpenAI meant to collapse customer acquisition costs.

This is basic business 101.

A single snapshot in time of API revenue means absolutely nothing for overall profitability

[Idrees Kamal @signedbykamal]

@klinkicz @ForwardEditor Just switch on Fast mode with Luna Max and you're still paying substantially less per token than Sol at any reasoning level

[Shreyas Doshi @shreyas]

It really is this simple:

[Idrees Kamal @signedbykamal]

Elon predicted Chinese models wouldn't reach Mythos level until Q1 2027.

Kimi K3 launched one month later and exceeds or is at par with US frontier models in most use cases.

The accelerations is real

[Elon Musk @elonmusk]

@jietang @teortaxesTex On benchmarks, yes, but as measured by true usefulness even Q1 would be very impressive.

Anthropic has rightly focused on maximizing useful intelligence, which does not show up in benchmarks, but definitely shows up in revenue.

[Idrees Kamal @signedbykamal]

“Vibe coding” is only 17 months old.

Since then:
• AI models have been banned by the U.S. govt
• Agents went from autocomplete to days of autonomous work
• Models learned to use computers better than humans

If you still don’t see the curve, it has already passed you.

[Andrej Karpathy @karpathy]

There's a new kind of coding I call "vibe coding", where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. It's possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper so I barely even touch the keyboard. I ask for the dumbest things like "decrease the padding on the sidebar by half" because I'm too lazy to find it. I "Accept All" always, I don't read the diffs anymore. When I get error messages I just copy paste them in with no comment, usually that fixes it. The code grows beyond my usual comprehension, I'd have to really read through it for a while. Sometimes the LLMs can't fix a bug so I just work around it or ask for random changes until it goes away. It's not too bad for throwaway weekend projects, but still quite amusing. I'm building a project or webapp, but it's not really coding - I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.

[Idrees Kamal @signedbykamal]

The real measure of AI psychosis should be if you use the words “permanent underclass” in your tweets

[Idrees Kamal @signedbykamal]

Between the Sam Altman ouster from OpenAI and the USG rug pulling Fable 5, the race for AGI is going to make for one HELL of a movie

[Idrees Kamal @signedbykamal]

@ajambrosino @thsottiaux @simpsoka is it possible to detach the codex browser from the chat? I often want to start a new chat (fresh context window) with the current browser window still active

right now i'm forced to start a new chat and log into what ever browser session i was in again

[Idrees Kamal @signedbykamal]

@MTSlive This is literally just one /goal prompt

[Idrees Kamal @signedbykamal]

This update is huge for enterprise environments, the vast majority of which use M365/Windows

We are seeing the building blocks of end to end agentic work being assembled in real time.

[OpenAI @OpenAI]

Windows users, this one’s for you.

Computer use now works on Windows, so Codex can take action on your Windows computer.

And with Windows support for Codex in the ChatGPT mobile app, you can start, review, and steer tasks on the go while work continues on your Windows machine.

An early experience, but we’re working on more ways to keep your work moving, wherever you are.

[Tyler @rezoundous]

Opus 4.8 is insane guys. It one shotted my session usage limit.

[Idrees Kamal @signedbykamal]

My boy aint neva let me down 🐐

[Idrees Kamal @signedbykamal]

Opus 4.8 seems like it would be genuinely helpful if it didn't chew through my limits after one task....

Yeah, sticking with Codex and 5.6

[Idrees Kamal @signedbykamal]

@fortelabs The human condition is such that freedom itself is just another form of imprisonment