Pinned Tweet
The new DeepSeek Flash release is WILD.
Not just super capable, but SUPER cheap.
I grinded half a day and it barely spend a $1.
The wildest part is, after using it for 3-4 days straight now, I honestly can't even tell a difference in quality from DeepSeek to GPT-6.1 Sol.
T 6 Astra and far behind Opus 5.5, though).
Still. Incredible where we got with open models.
Full breakdown 👇
Not just super capable, but SUPER cheap.
I grinded half a day and it barely spend a $1.
The wildest part is, after using it for 3-4 days straight now, I honestly can't even tell a difference in quality from DeepSeek to GPT-6.1 Sol.
T 6 Astra and far behind Opus 5.5, though).
Still. Incredible where we got with open models.
Full breakdown 👇
22
3
60
12,436
Simon Høiberg retweeted
While Europe is crashing and burning (literally), I'm setting up a new business in Singapore 🇸🇬
The process of setting up a business here is so smooth - very straightforward and simple.
I also get:
- Safest country in the world
- No riots in the streets
- Everyone's behaving g tax deal
- Business friendly
- Advanced & high functioning
If your store got looted or your car trashed and burned, just know there are places in the world where they figured it out.
The process of setting up a business here is so smooth - very straightforward and simple.
I also get:
- Safest country in the world
- No riots in the streets
- Everyone's behaving g tax deal
- Business friendly
- Advanced & high functioning
If your store got looted or your car trashed and burned, just know there are places in the world where they figured it out.
82
8
324
22,183
Either people complain about AI slop.
And if not, they complain it's not AI slop enough.
Hard to win 🤷
And if not, they complain it's not AI slop enough.
Hard to win 🤷
13
0
20
4,410
While Europe is crashing and burning (literally), I'm setting up a new business in Singapore 🇸🇬
The process of setting up a business here is so smooth - very straightforward and simple.
I also get:
- Safest country in the world
- No riots in the streets
- Everyone's behaving g tax deal
- Business friendly
- Advanced & high functioning
If your store got looted or your car trashed and burned, just know there are places in the world where they figured it out.
The process of setting up a business here is so smooth - very straightforward and simple.
I also get:
- Safest country in the world
- No riots in the streets
- Everyone's behaving g tax deal
- Business friendly
- Advanced & high functioning
If your store got looted or your car trashed and burned, just know there are places in the world where they figured it out.
82
8
324
22,183
Name a better combo than Switzerland 🇨🇭 and Singapore 🇸🇬
I'll wait.
I'll wait.
If you're a founder and a dad, you should live in 🇨🇭 Switzerland.
- Incredible nature
- Super safe and clean
- Low taxes
- Non-EU
- Great for business
- Perfect for families
The best place on earth for men to build real wealth (I'm not just talking about money).
4
0
21
1,714
Simon Høiberg retweeted
Back in Singapore 🇸🇬
CoWorking set up! Ready to work.
Time to set up a new business here! (Cool update coming soon 👀)
CoWorking set up! Ready to work.
Time to set up a new business here! (Cool update coming soon 👀)
19
1
64
4,870
Back in Singapore 🇸🇬
CoWorking set up! Ready to work.
Time to set up a new business here! (Cool update coming soon 👀)
CoWorking set up! Ready to work.
Time to set up a new business here! (Cool update coming soon 👀)
19
1
64
4,870
Should have been in Dubai. But replanned because of all the chaos in the gulf (again...) 😒
@marclou Hell yea! 🔥
In Dubai for the rest of the year, I'll be joining for sure.
0
0
3
701
This is what Europe is becoming. The new place where we ask "is it safe to come" - very sad.
Hi friends, looking for help from people in France!
I am supposed to fly next week to France for the Uneed residency in Nantes (I have a flight and everything)
How safe is it? My wife is a bit worried
9
2
75
11,188
I've always hated meetings. My whole career, I've tried to get rid of them.
Now I hold meetings with my AI team. And for the first time ever, I actually enjoy them.
We have a Telegram group with the whole team and a simple turn-taking system (kind of like raising your hand). I agent waits its turn, reads what the others said, and adds its take.
Our product specialist pitches a campaign, my YouTube strategist pushes back, and our accountant raises her hand with a problem nobody else saw.
No slides, no ego, and nobody getting hungry and wanting to wrap up.
Let me show you 👇
Now I hold meetings with my AI team. And for the first time ever, I actually enjoy them.
We have a Telegram group with the whole team and a simple turn-taking system (kind of like raising your hand). I agent waits its turn, reads what the others said, and adds its take.
Our product specialist pitches a campaign, my YouTube strategist pushes back, and our accountant raises her hand with a problem nobody else saw.
No slides, no ego, and nobody getting hungry and wanting to wrap up.
Let me show you 👇
20
0
36
4,871
📌 Most AI agent teams just produce slop. In my latest video, I share the 5 things I do to make mine actually perform 👇
youtu.be/-126yfiC2Mg
youtu.be/-126yfiC2Mg
1
1
5
883
I think we have 1-2 years left before the broader masses catch up on what you can actually do with AI today. Sell that gab until, there's some still time left to make cash.
But that's the timeline.
Company vision with decade-long horizon makes zero sense right now.
But that's the timeline.
Company vision with decade-long horizon makes zero sense right now.
Me thinking about my remaining moat
31
1
81
12,291
My AI agents tear each other's work apart before I ever see it.
Ask an agent to check its own work, and it'll tell you it's great. Models even rate their own writing higher than anyone else's.
So we don't let them grade their own homework.
Max (our YouTube strategist) reviews nd editor make. And he's NOT polite about it.
He's sent scripts back because a chapter was so busy covering itself with caveats that it forgot to be exciting. He's flagged a cover from our editor that implied an integration we don't even have.
What I catch now is taste, not typos and broken renders.
Here's how we set it up 👇
Ask an agent to check its own work, and it'll tell you it's great. Models even rate their own writing higher than anyone else's.
So we don't let them grade their own homework.
Max (our YouTube strategist) reviews nd editor make. And he's NOT polite about it.
He's sent scripts back because a chapter was so busy covering itself with caveats that it forgot to be exciting. He's flagged a cover from our editor that implied an integration we don't even have.
What I catch now is taste, not typos and broken renders.
Here's how we set it up 👇
31
2
39
3,918
📌 Most AI agent teams just produce slop. In my latest video, I share the 5 things I do to make mine actually perform 👇
youtu.be/-126yfiC2Mg
youtu.be/-126yfiC2Mg
0
0
3
664
No one actually believed this EU Inc was gonna be a hit, did they? 😂 Of course it's useless, most predictable outcome ever.
Europe’s founders need a European standard to compete globally from day one.
Europe’s leaders committed to delivering one. But their teams are negotiating away the very things that make it one.
There is still time to change course if Europe's Heads of State step in now.
31
5
96
22,326
Simon Høiberg retweeted
My AI team has 7 agents. Each one does exactly one job.
- John runs the operation and hands out the work
- Max grows my YouTube channel
- Elias writes the scripts
- Milo edits the videos
- Leo builds the products
- Claire handles support
- Liza handles bookkeeping
And they me model.
Codex runs the orchestration, the code and the operations.
Opus 5.5 does the writing, editing and creative work.
Qwen 3.8 runs on our own GPUs.
Most people give one agent everything and wonder why it falls apart.
One job each. The right model for each job.
I still bring the ideas and make the final call. The team does the rest.
Here's who does what 👇
- John runs the operation and hands out the work
- Max grows my YouTube channel
- Elias writes the scripts
- Milo edits the videos
- Leo builds the products
- Claire handles support
- Liza handles bookkeeping
And they me model.
Codex runs the orchestration, the code and the operations.
Opus 5.5 does the writing, editing and creative work.
Qwen 3.8 runs on our own GPUs.
Most people give one agent everything and wonder why it falls apart.
One job each. The right model for each job.
I still bring the ideas and make the final call. The team does the rest.
Here's who does what 👇
22
3
29
3,346
My AI team has 7 agents. Each one does exactly one job.
- John runs the operation and hands out the work
- Max grows my YouTube channel
- Elias writes the scripts
- Milo edits the videos
- Leo builds the products
- Claire handles support
- Liza handles bookkeeping
And they me model.
Codex runs the orchestration, the code and the operations.
Opus 5.5 does the writing, editing and creative work.
Qwen 3.8 runs on our own GPUs.
Most people give one agent everything and wonder why it falls apart.
One job each. The right model for each job.
I still bring the ideas and make the final call. The team does the rest.
Here's who does what 👇
- John runs the operation and hands out the work
- Max grows my YouTube channel
- Elias writes the scripts
- Milo edits the videos
- Leo builds the products
- Claire handles support
- Liza handles bookkeeping
And they me model.
Codex runs the orchestration, the code and the operations.
Opus 5.5 does the writing, editing and creative work.
Qwen 3.8 runs on our own GPUs.
Most people give one agent everything and wonder why it falls apart.
One job each. The right model for each job.
I still bring the ideas and make the final call. The team does the rest.
Here's who does what 👇
22
3
29
3,346
Simon Høiberg retweeted
Almost everyone starts with one AI agent and gives it everything.
- A support ticket
- A LinkedIn post
- Something weird on the server
- Then back to the support ticket
Context windows are huge now, so why not have a single agent do it all?
Well... more context doesn't mean d context rot. The longer the input gets, the less reliable the model gets (even when the answer is right there).
You'd never hire a single person to be your support team, your DevOps engineer AND your accountant.
So don't do it with your agents either.
Here's how we set ours up 👇
- A support ticket
- A LinkedIn post
- Something weird on the server
- Then back to the support ticket
Context windows are huge now, so why not have a single agent do it all?
Well... more context doesn't mean d context rot. The longer the input gets, the less reliable the model gets (even when the answer is right there).
You'd never hire a single person to be your support team, your DevOps engineer AND your accountant.
So don't do it with your agents either.
Here's how we set ours up 👇
31
2
51
4,886
Almost everyone starts with one AI agent and gives it everything.
- A support ticket
- A LinkedIn post
- Something weird on the server
- Then back to the support ticket
Context windows are huge now, so why not have a single agent do it all?
Well... more context doesn't mean d context rot. The longer the input gets, the less reliable the model gets (even when the answer is right there).
You'd never hire a single person to be your support team, your DevOps engineer AND your accountant.
So don't do it with your agents either.
Here's how we set ours up 👇
- A support ticket
- A LinkedIn post
- Something weird on the server
- Then back to the support ticket
Context windows are huge now, so why not have a single agent do it all?
Well... more context doesn't mean d context rot. The longer the input gets, the less reliable the model gets (even when the answer is right there).
You'd never hire a single person to be your support team, your DevOps engineer AND your accountant.
So don't do it with your agents either.
Here's how we set ours up 👇
31
2
51
4,886
📌 Most AI agent teams just produce slop. In my latest video, I share the 5 things I do to make mine actually perform 👇
youtu.be/-126yfiC2Mg
youtu.be/-126yfiC2Mg
2
0
7
713