Amine Azariz

Software engineer, photographer. 🇺🇸 / 🇫🇷
Currently @prime_fintech.

[Amine Azariz @AmineAzariz]

How to defeat a human : Be smarter.
How to defeat an AI (LLM) : "Are you sure?"

[Amine Azariz @AmineAzariz]

@eastdakota Question is : wth they were doing on Azure in the first place? how can CF not be the obvious choice when you're starting anything ...

[Amine Azariz @AmineAzariz]

"Lawyers will be the first to be replaced." You hear this a lot in the AI world, the narrative being: it's a job that relies heavily on written knowledge.
But when you see the benchmarks, Legal (and Finance) is where even the best models struggle the most.
It's not that simple.

[Amine Azariz @AmineAzariz]

For those who still read articles, this one is a must-read. Extremely interesting bits.
Note: The end is hilarious though and a bit delusional but still worth a read.

"When AI Builds Itself" – https://t.co/Zc9kfc2xMu

[Amine Azariz @AmineAzariz]

hey Japan, we have been waiting for you. The race is on...

[Amine Azariz @AmineAzariz]

The funniest thing Musk could do right now is to spinoff X from SpaceX and call it Twitter.

[Amine Azariz @AmineAzariz]

@ZouhairMaj Let’s hope this time it will go smoothly

[Amine Azariz @AmineAzariz]

Spacex IPO prospectus (2026) : mission is “to extend the light of consciousness to the stars”.

WeWork IPO prospectus (2009) : mission is “to elevate the world's consciousness”.

[Amine Azariz @AmineAzariz]

One of the biggest misconceptions about AI is that it’s just software, so each new version will be better than the previous one. ML is sometimes closer to chemistry than to software, improving is neither linear nor guaranteed. That’s why Sonnet 4.6 was worse than Sonnet 4.5.

[Amine Azariz @AmineAzariz]

In a couple of months, markets will realize that AI agents need stablecoins (crypto) to make payments. And people will go bananas, again.

[Amine Azariz @AmineAzariz]

Just heard "Why do AI articles always sound so robotic?" ... the irony.

[Amine Azariz @AmineAzariz]

Funny how with AI, the more it knows -the larger the context- the dumber it becomes. Vs. humans, the more they know about a problem the better they are at solving it.

[Amine Azariz @AmineAzariz]

Another way of creating a machine smarter than humans: make humans dumber. ChatGPT might actually succeed after all.

[Amine Azariz @AmineAzariz]

Human ingenuity for making a buck is limitless!

[Financial Times @FT]

France’s weather forecasting service has filed a police complaint after detecting anomalies in its temperature gauges at Paris-Charles de Gaulle airport, which coincided with a surge in well-timed bets on prediction market Polymarket. https://t.co/r0md98xLRE

[Amine Azariz @AmineAzariz]

Future take: superstar developers will be able to have their skills packaged as paid agents/plugins.

Companies can hire/install 'you' in their AI stack and pay you a small salary, but an uncapped one depending on how good you are and how many are using your skills.

[Amine Azariz @AmineAzariz]

I've been using Gemini CLI (in parallel with Claude) for a bit now to see what's out there on different kinds of projects—large and small, complex and simple.

Takeaway: We're still in a world where nothing comes close to Claude Code.

[Amine Azariz @AmineAzariz]

the strait of Schrödinger

[Amine Azariz @AmineAzariz]

And not even talking about techie things, like Claude Code channels, scheduled prompts, /loop, remote control and claude code cloud. Or things like the Claude Advisor setup: https://t.co/qfFPbU3SzD

Anyways... "Become at one with your weapon." has never been so current.

[Amine Azariz @AmineAzariz]

You actually don't need anything more than Claude. Stop buying mac minis for openclaw, etc. Spend some time really understanding the tools you already have/pay for: check Claude Cowork with Dispatch and Claude Managed Agents ... and thank me later.

[Amine Azariz @AmineAzariz]

The next financial crisis/bailout won’t be a bank. It’ll be an AI company that became too big to fail.