Kapil Paliwal

Working at @outrank_so $250k/m,
1 (acq.)+8 more

[Kapil Paliwal @kapilsuham]

I am happy to announce my 10 startup 🥳

Say hello to SupportGPT – the AI-powered support agent that answers your customers instantly, 24/7.

No code. No hassle. Just smarter support. 💬✨

Try it now for free → https://t.co/LAmjYf7ArJ

[Kapil Paliwal @kapilsuham]

@DanielLockyer Brother I can relate to you 😭

[Kapil Paliwal @kapilsuham]

The first & most exposed profession to ai is content writing but still we see ppl making $ out of it

I know I know that 80% of content we generally read these days are in a way cooked with ai but still we can't ignore the fact that ppl those who qre still making money with content writing. 2026 are making 4x money vs earlier for same work

Thanks to ai

[Kapil Paliwal @kapilsuham]

Absolutely mind-blowing that a Chinese open-weights model like GLM-5.2—trained without Nvidia GPU—is legitimately flexing on Claude Opus 4.8 in open coding harnesses. 🤯

How did they even pull this off?!"

[Submitmysaas @submitmysaas]

🏆 Here's last week's podium, congratulations

🥇 The Tool Empire
🥈 Vectorize Image

[Kapil Paliwal @kapilsuham]

I have an mobile app out of my 20+ products which is only used by meninthe whole world 🌍

No users no mrr

Still proud of myself on that one

[Kapil Paliwal @kapilsuham]

@legends_loot @omarvvvr Launch on https://t.co/y0e8rBFO31

[Kapil Paliwal @kapilsuham]

The first & most exposed profession to ai is content writing but still we see ppl making $ out of it

I know I know that 80% of content we generally read these days are in a way cooked with ai but still we can't ignore the fact that ppl those who qre still making money with content writing. 2026 are making 4x money vs earlier for same work

Thanks to ai

[Daniel @growing_daniel]

Why is AI writing still so bad

[Kapil Paliwal @kapilsuham]

Absolutely mind-blowing that a Chinese open-weights model like GLM-5.2—trained without Nvidia GPU—is legitimately flexing on Claude Opus 4.8 in open coding harnesses. 🤯

How did they even pull this off?!"

[Kapil Paliwal @kapilsuham]

@ThomasRayvol Launch on https://t.co/y0e8rBFO31

[Kapil Paliwal @kapilsuham]

I have an mobile app out of my 20+ products which is only used by meninthe whole world 🌍

No users no mrr

Still proud of myself on that one

[Submitmysaas @submitmysaas]

🏆 Here's last week's podium, congratulations 🙌

🥇 Funboxie
🥈 @astic__ai
🥉 @hintder_ai

[Kapil Paliwal @kapilsuham]

It's easy to make 2k$/month remotely if you can grind for 2 yr straight

That's all it needs

[Kapil Paliwal @kapilsuham]

@DanKulkov Close to 2.5k mrr

[Kapil Paliwal @kapilsuham]

Yep yep we made it

Our performance is improving

Let's gooooooo

[Kapil Paliwal @kapilsuham]

It's easy to make 2k$/month remotely if you can grind for 2 yr straight

That's all it needs

[Kapil Paliwal @kapilsuham]

@stemonteduro Becz he also recently posted a theory post - https://t.co/d1FvaafWLv

[Tibo @tibo_maker]

everyone thinks the US government bullied Anthropic into killing Fable 5

I think Anthropic wanted it

here's my 5-part theory 👇

···
part 1: Fable 5 was a hit

it's the most powerful model Anthropic has ever put in public hands, and people jumped on it fast

the problem is Anthropic has been compute-starved for the better part of a year. rate limits, peak-hour throttling, outages. they admitted it themselves in may

and Fable 5 costs double what Opus does

a hit product is a good problem for most companies. for a compute-starved one, it's just a bigger bill

···
part 2: Amazon found the jailbreak

not some random researcher. Amazon 🤔

the company that's both a major Anthropic investor and one of the clouds running its models

they found the flaw, took it straight to the government, and the government told Anthropic to fix it or pull it

···
part 3: Anthropic said no

they called the jailbreak narrow and said other models like OpenAI's GPT-5.5 can do the same thing

they probably knew refusing would force a shutdown, which let them save face

so the government hit Fable 5 with an export control and Anthropic had to kill it for everyone

a company refusing a safety request right after filing for an IPO is not normal behavior

unless losing the model was never really a loss 🧐

···
part 4: they were already rationing it

before any jailbreak, before any of this, Anthropic had announced Fable 5 was coming off its subscription plans on june 23

their words: they'd put it back when "sufficient capacity allows"
plain version: they were bundling their most expensive model into existing plans at no extra charge, which is brutal to serve at scale

the shutdown just took it the rest of the way, eleven days before Anthropic was going to limit it themselves

···
part 5: the victim becomes the hero

"we can't serve our own flagship" is a humiliating headline
"the government is censoring us" is a different story

and remember, days before launch Anthropic was publicly calling for a "brake pedal" on frontier AI and saying labs should have the option to slow down. then they shipped their most powerful model ever

now they get to look principled instead of broke. good guys branding again

···
so here's the whole thing: Anthropic was going to ration Fable 5 anyway
the government handed them a way to do it that makes them look brave instead of capacity-strapped

this whole thing can be looked at as Anthropic buying itself time on compute, without ever having to admit it couldn't keep up with its own growth

they didn't fight it because they didn't need to

I'm not saying they staged it. I'm saying it worked out a little too cleanly 🤷🏻

[Kapil Paliwal @kapilsuham]

@jackfriks I have made over 20 products still waiting for it