The Tools Big Tech Hides

AI tools reviewed & compared — no fluff, no hype. New here → check pinned

[The Tools Big Tech Hides @_AIThatSlaps_]

I GAVE AN AI $50 AND TOLD IT TO PAY FOR ITSELF OR DIE.

A few hours later, it was down to $11.

So it changed the strategy on its own.

Crypto out.
Weather markets in.
I closed the laptop.

It kept scanning, trading and managing its own bankroll.

48 hours later:

$50 → $5,191.

227 trades.
219,000 markets scanned.
$200 subscription paid from the profits.

The part I can’t get over is the dip to $11.
It was basically dead.
Then it changed what it was doing and kept going.

Now I want to see what happens after a week.

[The Tools Big Tech Hides @_AIThatSlaps_]

MOST PEOPLE ARE STILL USING AI LIKE A CHATBOX.

That’s about to look ancient.

Grok Bot can keep watching, reasoning and working after you close the laptop.

The real AI race is moving from intelligence to execution.

I broke down the whole shift here ↓ https://t.co/bvWs9saYE9

[The Tools Big Tech Hides @_AIThatSlaps_]

I GAVE AN AI $50 AND TOLD IT TO PAY FOR ITSELF OR DIE.

A few hours later, it was down to $11.

So it changed the strategy on its own.

Crypto out.
Weather markets in.
I closed the laptop.

It kept scanning, trading and managing its own bankroll.

48 hours later:

$50 → $5,191.

227 trades.
219,000 markets scanned.
$200 subscription paid from the profits.

The part I can’t get over is the dip to $11.
It was basically dead.
Then it changed what it was doing and kept going.

Now I want to see what happens after a week.

[The Tools Big Tech Hides @_AIThatSlaps_]

@0xWast3 this is the part that gets me.
they’re both just doing the work

[The Tools Big Tech Hides @_AIThatSlaps_]

ANTHROPIC JUST SHOWED WHY AI AGENCIES ARE ABOUT TO GET WEIRD.

One system.

10 clients.

Same 3 models running all of them.

Haiku decides what matters.
Sonnet writes.
Opus sits at the gate and decides what actually reaches the client.
The crazy part is client #11.
No new hire.
No second machine.
No new workflow.

You add one paragraph describing the client…

and the cost goes up by about $2.45/day.
That’s it.
The expensive part was building the system once.

After that, every new client is basically another block in the file.

10 accounts: ~$735/month.

A traditional agency could charge ~$20,000 for the same number of clients.

This is the part people still underestimate about AI.

It doesn’t just make work cheaper.

It makes the next customer almost free.

[The Tools Big Tech Hides @_AIThatSlaps_]

ANTHROPIC JUST SHOWED WHY AI AGENCIES ARE ABOUT TO GET WEIRD.

One system.

10 clients.

Same 3 models running all of them.

Haiku decides what matters.
Sonnet writes.
Opus sits at the gate and decides what actually reaches the client.
The crazy part is client #11.
No new hire.
No second machine.
No new workflow.

You add one paragraph describing the client…

and the cost goes up by about $2.45/day.
That’s it.
The expensive part was building the system once.

After that, every new client is basically another block in the file.

10 accounts: ~$735/month.

A traditional agency could charge ~$20,000 for the same number of clients.

This is the part people still underestimate about AI.

It doesn’t just make work cheaper.

It makes the next customer almost free.

[wast3 @0xWast3]

https://t.co/j9thHBFgdB

[The Tools Big Tech Hides @_AIThatSlaps_]

@0xWast3 Making disagreement a feature instead of a failure is the clever part

[The Tools Big Tech Hides @_AIThatSlaps_]

@Atenov_D being a student suddenly has perks again lol

[The Tools Big Tech Hides @_AIThatSlaps_]

300 AI AGENTS CAN GIVE YOU 300 PERFECT ANSWERS AND STILL MISS THE ACTUAL POINT.

Because the useful part isn’t always inside the answers.

It’s between them.

10 findings = manageable.

100 findings = 4,950 possible relationships.

300 agents dumping notes into a folder isn’t research.

It’s a very fast mess.

The real trick is making them return the connections:

who depends on who
what contradicts what
what caused what
what keeps showing up together
The video makes this painfully obvious.
The nodes grow slowly.
The edges explode.

Everyone is obsessed with making agents fan out.

The merge is where the intelligence actually happens.

[slash1s @slash1sol]

https://t.co/fYpaaiH1dP

[The Tools Big Tech Hides @_AIThatSlaps_]

@TeddyinMedia Yeah. The 24/7 part is way bigger than the cost cut

[The Tools Big Tech Hides @_AIThatSlaps_]

WALL STREET SHOULD PROBABLY BE PAYING ATTENTION TO THIS.

Someone built a 6-agent AI research desk that runs all night while you sleep.

One reads filings.
One listens to earnings.
One watches sectors.
One tracks insiders.
One monitors chatter.
Then a “Chief of Staff” agent reads everything, throws out the noise, and builds the morning brief.
The dashboard literally shows them handing work to each other in real time.
$294,000/year research stack → ~$2,400/year.
122x cheaper.

And the weirdest part?

There’s no analyst sitting there at 2am.

The whole desk is just… running.

[cvxv666 @antpalkin]

https://t.co/LiPq7x0lPK

[The Tools Big Tech Hides @_AIThatSlaps_]

@Caarat1 Exactly. Cheap is nice.

Never sleeping is the actual advantage.

[The Tools Big Tech Hides @_AIThatSlaps_]

@minimachine_ Yeah, that’s the part that feels the most unreal

[The Tools Big Tech Hides @_AIThatSlaps_]

GROK IS NOW RUNNING A WALL STREET TRADING FLOOR.

6 AI agents.

TAPE.
QUANT.
MACRO.
RISK.
FLOW.
PM.
The morning shift starts at 4AM.
The owner isn’t even there.
Each agent has its own computer, browser and terminal, while a Chief of Staff routes the work between them.
In the video the desk runs 125 orders, gets 130 fills, touches +$173K…
and still ends the session down $8K.

No fake perfect AI demo.

Just a trading floor running while the laptop is closed.

This used to take a team of people on six-figure salaries.

Now it takes an evening to set up.

[cvxv666 @antpalkin]

https://t.co/LiPq7x0lPK

[The Tools Big Tech Hides @_AIThatSlaps_]

@0xWast3 The “only interrupts you to disagree with you” part is actually the most interesting bit

[The Tools Big Tech Hides @_AIThatSlaps_]

WALL STREET SHOULD PROBABLY BE PAYING ATTENTION TO THIS.

Someone built a 6-agent AI research desk that runs all night while you sleep.

One reads filings.
One listens to earnings.
One watches sectors.
One tracks insiders.
One monitors chatter.
Then a “Chief of Staff” agent reads everything, throws out the noise, and builds the morning brief.
The dashboard literally shows them handing work to each other in real time.
$294,000/year research stack → ~$2,400/year.
122x cheaper.

And the weirdest part?

There’s no analyst sitting there at 2am.

The whole desk is just… running.

[The Tools Big Tech Hides @_AIThatSlaps_]

@zefirium $5 a day just for checking in is kinda wild.

AI pricing is turning into a game.

[The Tools Big Tech Hides @_AIThatSlaps_]

@minimachine_ And it’s only going to get cheaper from here.

[The Tools Big Tech Hides @_AIThatSlaps_]

THIS SHOULD NOT BE POSSIBLE ON A $300 GPU.

Qwen 3.8 27B is running locally on an RTX 4060.

8GB VRAM.

64K context.
14.6GB on disk.
~150 tok/s prefill.
~5 tok/s decode.
And somehow it stays inside those 8GB.
No 4090.
No $10,000 workstation.
No cloud bill quietly eating your wallet.
Just a 4060 sitting at ~97% VRAM running a 27B model on localhost.

That’s the crazy part.

We’re not just getting better models.
We’re getting insanely capable models on hardware millions of people already own.
Local AI is moving way faster than most people realize.

[CyrilXBT @cyrilXBT]

https://t.co/EqoOhLWuDa

[The Tools Big Tech Hides @_AIThatSlaps_]

@TeddyinMedia @bcherny At some point, more instructions stop helping.

They just get in the model’s way.