John · Sep 12, 2026 · 7:33 PM UTC

John

267Media

Pinned Tweet

John@hottubjohnSep 12

The more I learn, the more there is to learn.

John · Oct 18, 2026 · 1:42 AM UTC

John@hottubjohnOct 18

#̢̛ͥ̀̽̽ͨ̌̓͆̇̀̽ͦͨͫͯ̏̎̇̈̚͜҉̕͢͏͡҉̸̶̛͢͡҉̵̨̭̳̘̹̫͔͖̜̠͖̫̲̖̱̤̻̖͉͢͞͡ͅͅ͏̸̷̢͑̊̀̐ͣͭͭ̉͂ͨ̽͒̊̎͛ͪ̊̎ͩͧ̀͝͠͠͏̸̵̵̷̵

John · Oct 16, 2026 · 5:35 PM UTC

John@hottubjohnOct 16

it has to be the best because it is the best

John · Oct 16, 2026 · 1:39 AM UTC

John@hottubjohnOct 16

false positives, cool. false negatives, no.

John · Oct 12, 2026 · 6:22 PM UTC

John@hottubjohnOct 12

Small shops can miss payout shortages and excess fees. I used @AutoClawAIer to build LedgerLens from a PRD to audit them. Fictional demo: 7/7 planted leak classes found, $816.25 flagged, zero false positives. Demo: ledgerlens-kappa.vercel.app

AutoClaw verification beside the LedgerLens fictional-shop dashboard. The verification reports a data-derived timestamp and 13 independent checks passing. The dashboard shows 7/7 leak classes, zero false positives, $816.25 leaked, and $40,465.25 gross merchandise value.

ALT AutoClaw verification beside the LedgerLens fictional-shop dashboard. The verification reports a data-derived timestamp and 13 independent checks passing. The dashboard shows 7/7 leak classes, zero false positives, $816.25 leaked, and $40,465.25 gross merchandise value.

John · Oct 11, 2026 · 6:02 PM UTC

John@hottubjohnOct 11

the best model i’ve recently tested is insane. it absolutely rivals DeepSeek. it’s from @Zai_org & it takes the ‘Flash’ gen to the neXt level. trust. #zai

John · Oct 11, 2026 · 4:31 AM UTC

John@hottubjohnOct 11

tern from @stencil_labs is truly a wonder. installed it & haven't stopped using it since. it's my daily driver. Thanks to the team! 🎉

John · Oct 4, 2026 · 11:07 PM UTC

John@hottubjohnOct 4

huge flex from @ManusAI manus.im/blog/introducing-ma…

Introducing Manus Flex: Use your own API key in the Manus workspace

Manus is launching Manus Flex, a new module that lets users power Manus with their own API key from a supported inference provider.

manus.im

John · Sep 30, 2026 · 4:40 PM UTC

John@hottubjohnSep 30

GLM will win.

Andrew Case

@attrcSep 29

I am very confused why Anthropic wrote a blog post advertising for cybersecurity teams to switch to GLM!?!? anthropic.com/research/glm-5…

GLM-5.3 and the spread of advanced cyber capabilities

GLM-5.3 can autonomously build end-to-end cyber exploits, but unlike other frontier models, it was released without meaningful safeguards to limit misuse.

anthropic.com