Basil Chatha

1.1K posts

Basil Chatha banner
Basil Chatha

Basil Chatha

@realbasilchatha

alleged cyber criminal | prev PM @creditkarma, CoS @jspujji

San Francisco, CA Katılım Kasım 2020
1.8K Takip Edilen498 Takipçiler
SMB Attorney
SMB Attorney@SMB_Attorney·
Hear me out: We’re putting hundreds of people into a multi-million-dollar machine, navigating it 30,000 feet in the air at 500+ mph, powered by thousands of gallons of fuel, maintained by experts, and coordinated globally… …and we’re surprised it can’t be done for low cost?
English
35
7
157
19.5K
Bryce Roberts
Bryce Roberts@bryce·
At a recent Anthropic user conference they noted that for every dollar spent on their software, customer spent $8-$12 on services. You’re not bullish enough on AI services.
English
17
19
257
82.6K
Ayman Al-Abdullah 🧱
Ayman Al-Abdullah 🧱@aymanalabdul·
Getting requests from clients for real AI implementation partners. All I'm finding are Vibe Code Bros or Zapier shops. I want firms that: • Diagnose the actual business problem • Bring PMs + Product + AI talent • Build + integrate into real workflows • Care about security and stability • Ship and iterate Who’s best in the world at this?
English
254
9
379
48K
Basil Chatha
Basil Chatha@realbasilchatha·
@MediaKing Do you think the TBPN deal was a bad outcome for those guys?
English
0
0
1
89
Matt Paulson
Matt Paulson@MediaKing·
Hot take: you probably shouldn't sell your media company. Every acquisition in financial media has failed. Media = great cash flow, terrible exit multiples. The math only works if you hold forever.
English
22
3
130
10.9K
Basil Chatha
Basil Chatha@realbasilchatha·
@MediaKing You should be running multiple Claude work trees in parallel so you always have something to attend to!
English
0
0
1
43
Matt Paulson
Matt Paulson@MediaKing·
When I was a kid, I would wait impatiently several minutes for my code to compile. Now as an adult, I impatiently wait several minutes for Claude Code to write the thing, then it compiles rather quickly. Same problem, 30 years apart.
English
9
1
36
2.7K
@jason
@jason@Jason·
We started an AI founder twitter group... reply with "I'm in" if you're a founder and want to be added
English
10.9K
136
4.6K
900.2K
Rusty ⚡️: Solar Powered ☀️
> be pregnant > buy induction cooktop > "cleaner than gas, no fumes, safer for the baby" > stand at normal cooking distance every day > baby's developing nervous system now exceeds public EMF limits entirely > stove passed IEC 62233 certification > test was done 1 foot away from the cooktop > you don't cook from 1 foot away > CERTIFIED SAFE
Rusty ⚡️: Solar Powered ☀️@ze_rusty

The Electricity bill is the cheap part. A Swiss research group tested 13 professional induction cooktops & found that most EXCEEDED even the already-high 1998 ICNIRP maximum exposure limits when a person stood close to the stove. For the worst-case compliant device: current density in tissue exceeded ICNIRP limits by up to 24 dB.. A FACTOR OF 16. The brain tissue of young children: overexposed by a factor of 2 The CNS of a fetus: can exceed general public limits entirely if the mother cooks at normal working distance. And this happened on devices that had PASSED the official safety test. Why? Because the IEC 62233 compliance standard, the test that certifies your stove as "SAFE", measures the AC magnetic field at 300mm (1 foot) from the cabinet edge. You don't cook from 1 foot away. Your abdomen is at 0mm. Your hands are at 50mm. Your child's head is at cooktop height. At 300mm: COMPLIANT At 0mm: 16x OVER THE LIMIT The test is designed to PASS not to protect. And that's before you consider that the ICNIRP limits themselves are already deeply inadequate. ICNIRP limits are set to prevent ACUTE effects.. enough induced current to directly stimulate a nerve. They don't account for chronic, non-thermal biological effects. Reba Goodman & Martin Blank showed magnetic field biological effects occur at 14 ORDERS OF MAGNITUDE below the thermal threshold ICNIRP uses as its baseline. The WHO IARC classified ELF AC magnetic fields.. the exact type induction stoves emit at 20 kHz, as a Group 2B possible carcinogen in 2002. ICNIRP's response in 2010 was to RELAX the reference levels. So when the study says your stove exceeds limits by a factor of 16.. That's 16x over a number that's already orders of magnitude too permissive for chronic daily exposure. You're not cooking on a stove. You're standing in front of an industrial ELF AC magnetic field generator. EVERY SINGLE DAY. But sure. The electricity bill is the part to worry about.

English
68
707
6.2K
706.6K
Basil Chatha
Basil Chatha@realbasilchatha·
6/ On why AI-generated code is less secure today: Models optimize for passing tests. Not security. Not quality. The reward function is laser-focused on green tests. Everything else? Ignored. RL optimizes for what you measure. Measure the wrong thing and you get insecure, ugly code that technically works, but only superficially.
English
1
0
2
77
Basil Chatha
Basil Chatha@realbasilchatha·
A few weeks ago, I hosted a fireside chat with the founders of Cognition ($898M), Semgrep ($204M), Factory ($70M), and Composio ($29M) on the future of agentic engineering. Here’s the reality: AI doesn’t make dev teams better. It reveals whether your codebase is actually agent-ready… or completely breaks under pressure. Some of what we covered + full episode in the comments ↓
Basil Chatha tweet media
English
2
0
4
149
khushi
khushi@khushkhushkhush·
there’s a young woman in tech (very active on x) that i’ve looked up to for the better part of 3 years just very delightful internet presence, heard nothing but great things from other founders initially, clearly a titan of the industry met her irl a few months ago and she was so unimaginably rude to me that it encouraged me to forge my own path and work for myself. minutes before meeting her i would have set everything aside for her to be my boss such a shame, but i look up to her and think of her fondly nonetheless — would not have found my calling otherwise
English
124
12
1.7K
336.5K
Hubert Thieblot
Hubert Thieblot@hthieblot·
pitch me your company in 1 word.
English
2.5K
14
1.1K
287.2K
Basil Chatha retweetledi
Michelle Fang 🌁
Michelle Fang 🌁@michelleefang·
Thursday 4/16 (cont'd) ‣ Getting started with Vibe Coding luma.com/sliyfinb @pebble_bed ‣ Founder Real Talk @ EF - How to Find Product-Market Fit luma.com/frt-apr2026 ‣ SF Poker Club: Poker Night (for Dummies) luma.com/vr0xiyc1 @realbasilchatha ‣ $100K AI Startup Pitch-Off @ Chargebee Beelieve '26 luma.com/beelieve ‣ AVs & the City: DRIVE AI Demo Day luma.com/901dhzn3 ‣ The Agentic Enterprise Platform — ThinkingAI Launch with MiniMax luma.com/ip1wjj8i @dasha_shunina ‣ ML Talks on Tap luma.com/yqh9u09m @adelwu_ @creet_z ‣ Bricks and Sips Happy Hour luma.com/b4rjuhcv @Vizcom_ai
English
1
1
5
1.6K
Basil Chatha
Basil Chatha@realbasilchatha·
@braelyn_ai Nice read! Although I do think the timeline will be much, much more compressed. There was no wide-adoption of the internet so information didn't travel as fast. Also, we have A LOT more people who have the fundamental CS/eng backgrounds to expedite that timeline @braelyn_ai
English
1
0
0
15
Braelyn ⛓️
Braelyn ⛓️@braelyn_ai·
codegen is still very far from its Gang of Four moment 1957 - compilers invented 1959 - today's equivalent 1968 - Dijkstra says GOTO statements are harmful 1994 - Gang of Four releases the concept of software design patterns 37 years between invention and best practices we're only 5.4% into the tool adoption arc you can stop dooming, its still early
English
10
4
41
4.8K