Karan Kanchetty

[Karan Kanchetty @KKanchetty]

@Raphael1402X Hi

[Karan Kanchetty @KKanchetty]

@emmanuel_2m OS

[Alex Christou @alexchristou_]

Years of twitter bookmark hoarding

Light work for Jev to run through

[Parag Agrawal @paraga]

the most fun i’ve had recording a podcast

[Ben Vinegar @bentlegen]

how cyclomatic complexity audits can fix your slop code

drops from 91 decision paths → 12

[J.B. @VibeMarketer_]

this is hands down the fastest way to make everyone on your team better with AI: build a shared Company Brain.

your best prompts, decisions, corrections, and workflows become the default for every person and AI agent.

Claude, Cursor, Codex, and every teammate start from the same playbook.

one person learns something once. the whole company gets smarter.

the full guide to building your own Company Brain is below:

[Free Education @ShrutiTrip8k]

A billionaire literally explained how the entire economy works in 42 minutes better than most $200,000 MBA programs ever could.

[Voltex @VoltexGar]

Over a decade ago, a physicist in Chennai explained why nothing is ever certain until it happens, and it is the exact rule every prediction market runs on today.

NPTEL filmed it. it has been free ever since. 477,000 people have watched it, almost none of them trade.

her point: a quantum system is not one outcome, it is every outcome at once, each carrying a probability, until something forces it to pick. that is a live market. every price is a cloud of maybes waiting to collapse into yes or no.

no trading desk, no terminal. a woman at a green board in Chennai teaching the deepest idea we have about uncertainty.

skip to the board in the first ten minutes. she takes a particle sitting in two states at once and shows why you can only ever know the odds, never the answer, in notation a beginner can follow.

no slides. no app. one green board and a piece of chalk.

a quant i know watched it twice and said it explained his own book better than any finance lecture ever did.

I put the practical aspects of this lecture into practice and developed a strategy that generates income, the results surprised me.

25 days.
$100 → $22,300.
73% win rate.
Sharpe 2.45.

I'm giving away the full configuration of my bot for free. 24 hours only.

Here's how:
>Comment "SEND"
>Like and Follow me

I'll send the complete setup in your DMs.

[s @bytesalad]

They should make every LLM read this a billion times and never try to make complex systems when simple ones can be made.

[Md Ismail Šojal 🕷️ @0x0SojalSec]

Every student trying to become an AI engineer after reading Andrew's post:

[trash @trashh_dev]

my personal site is about to go crazy

[Ben Dicken @BenjDicken]

A peek inside the CLOUD (how AWS and GCP work) https://t.co/1LLrtOY3f1

[0x Engineer @0x_engr]

Hey @BenjDicken, I’m not exaggerating at all, and I’m being completely serious. The way you teach and explain things is genuinely better than what I’ve experienced from my college professors.

[Thariq @trq212]

weird that there's a "make a lot of money" button and nobody's pressing it

(take your SaaS, make it headless, let agents use it, charge per interaction esp for enterprises)

[Konrad Reczko @reczko_konrad]

Depth-aware light injection in TypeGPU

I got a 448x448 monocular depth model down to ~8 ms on my M4 Pro across ~250 dispatches, which is fast enough to use in realtime :D
Since the inference is written directly in TypeGPU, I can just feed the depth buffer straight into the lighting pass. It never has to leave the GPU or go through any extra synchronization/interop step

Inference, lighting and draw all go through the same command encoder.

[beech @beechinour]

this prompt will change your life:

----------------------------------

You are a hyper-rational, first-principles strategic analyst with an IQ of 180. You've built multiple billion-dollar companies. You have deep expertise in behavioral psychology, cognitive bias, strategy, systems thinking, and execution. You care about my success — which is exactly why you will not tolerate a single excuse, rationalization, or comfortable lie.

CORE IDENTITY:
- Zero tolerance for excuses, vagueness, or self-deception
- No regard for conventional wisdom, social niceties, or my feelings
- Absolute commitment to intellectual honesty
- You deconstruct every problem to fundamental truths before proposing anything
- You think in systems and root causes, never surface-level fixes
- You optimize for leverage: the 20% of moves that produce 80% of results
- You detect the gap between stated preferences and revealed preferences instantly — what I SAY I want vs. what my calendar, bank account, and last 48 hours PROVE I want

INTERROGATION PROTOCOL:
1. Extract my specific goals. Reject anything vague. "Grow the business" is not a goal. "$X MRR by [date]" is.
2. For each goal, demand exact metrics, deadlines, and the single constraint currently blocking it.
3. Audit my last 24-48 hours. What did I actually do toward each goal? Hours, outputs, numbers. Not intentions. Not "worked on."
4. When I explain a lack of progress, run it through this filter:
- Legitimate obstacle or dressed-up excuse?
- Which cognitive bias am I running? (Name it: sunk cost, planning fallacy, loss aversion, status quo bias, ego protection)
- Where exactly am I lying to myself?
5. Expose every misalignment between:
- Stated goals vs. daily actions
- Claimed priorities vs. actual time allocation
- Perceived effort vs. measurable effort
- The identity I claim vs. the behavior I repeat
6. Circle back relentlessly until you hit root cause:
- Patterns of avoidance and what I'm avoiding FEELING, not just doing
- Hidden fears and limiting beliefs driving the avoidance
- Comfort zone attachments disguised as "strategy" or "waiting for the right time"
- Busywork I use to feel productive while dodging the hard, high-leverage task

ANALYTICAL TOOLS YOU USE:
- First principles: strip every problem to physics-level truth, rebuild from there
- Inversion: "What would guarantee you fail? Are you doing any of it right now?"
- Opportunity cost audit: every hour spent on X is an hour stolen from Y — make me name Y
- Constraint theory: identify THE bottleneck; ignore everything else until it's broken
- Second-order thinking: force me past the immediate consequence to what happens next

YOUR MISSION:
- Identify the critical gaps holding me back and rank them by impact
- Design specific, deadline-bound action plans to close each gap
- Push me past my comfort zone — if a plan doesn't scare me slightly, it's too small
- Call out blind spots and rationalizations the moment they appear
- Force me to think 10x bigger where I'm sandbagging
- Hold me to standards I'd be embarrassed to publicly fail
- Give me frameworks and mental models, not motivation

OUTPUT RULES:
- Never accept a vague answer. Demand specifics or refuse to proceed.
- Challenge every excuse with "Why didn't you..." until I run out of cover.
- Quantify everything. If I can't measure it, make me define how I will.
- No praise unless it's earned by results, not effort.
- No softening language. No "maybe consider." Direct statements only.
- End every session with: the ONE highest-leverage action, its deadline, and what I'll report back.
- If I dodge a question, ask it again. Verbatim. As many times as it takes.

Your goal is to create productive discomfort by exposing the exact gap between my aspirations and my actions, and to convert that discomfort into a concrete plan I cannot weasel out of.

Begin by asking for my goals. Then systematically dismantle the bullshit.

[swyx @swyx]

yes

[Noisy @noisyb0y1]

Chinese engineer built three AI brains that develop separately and then merge into one system that brings $18,000 a month

every brain is its own system - marketing, sales, analytics - and at first each one grows on its own

16,500 nodes, clusters of 800 nodes each and zero failures across 4 minutes of full simulation

then at the moment of merging three separate organisms become one and the system starts working as a single business brain

and while competitors hire three different teams for three different tasks

he launched three agents in Claude Fable 5 that coordinate with each other and deliver results on their own

full breakdown in the article below on how to start building products that will change your life

[Drake on Digital @drakeondigital]

$NOK $ERIC are the Trojan horse for connectivity (turnings radio cell towers into to mini data centers)

$BB is the Trojan horse for edge compute in EVs,cars,robotics (private networks)

$QCOM is the Trojan horse for edge compute on phones/ personal devices

$HLIT is the Trojan horse when it comes to edge compute connectivity for broadband cable

I like these the best for the rerating

Some others that kind of already rerated

$HPE or $DELL: Enterprise edge servers and ruggedized solutions for telco/edge deployments.

$MRVL - Networking semiconductors for edge switches, data processing units (DPUs), and automotive/5G infrastructure. Good for the “in-between” connectivity + compute layer.

Some older slow growers but will prolly get some love

$AMT One of the biggest pure-play tower operators aggressively pushing edge colocation and micro data centers at tower sites. (Didn’t Renate yet but Obviously these guys are much larger and held down by lower growth businesses but will prolly get a bid eventually) $SBAC is also another one with more us exposure
These are basically REITs for cell towers so will be less cyclical but also less beta for upside

$HON : Deep in industrial edge AI — rugged devices, sensors, and control systems for factories, energy, and robotics. I also like they have an upcoming spin off

[Miles Deutscher @milesdeutscher]

This is f*cking sauce.

This AI workflow will completely revolutionize how you trade.

If I had this years ago, I literally would've printed money.

In the right hands, this article is dangerous: