Karl Mehta Profile picture
Jul 21 15 tweets 5 min read Read on X
BREAKING: Norway's $2 trillion wealth fund ran a 12-month AI experiment.

They gave Claude access to their entire investment workflow.

Result: 213,000 hours saved. 20% productivity boost.

But what they found hiding in the data changed everything:

A Thread 🧵 Image
Image
Meet Nicolai Tangen, CEO of NBIM - the world's largest sovereign wealth fund.

700 employees managing $2 trillion in assets.

In 2022, he made a strategic decision that would reshape their entire operation.

But first, he had to address significant organizational resistance... Image
Tangen began systematically advocating for AI adoption across his 670-person team.

The challenge? Most investment professionals viewed AI as potentially disruptive to established workflows.

Traditional analysts were spending days on tasks that seemed impossible to automate.

Then Claude was introduced...
The first major breakthrough came with their Snowflake data warehouse integration.

Portfolio managers could suddenly query complex datasets using natural language.

What previously required technical SQL expertise now took seconds.

But this was just the initial phase... Image
Next, they implemented automated earnings call analysis.

Claude could process hours of executive commentary and extract key insights efficiently.

Risk managers were analyzing significantly more companies in the same timeframe.

Meanwhile, something important was happening behind the scenes...
NBIM monitors news for 9,000 companies across 16 languages.

Before Claude: Teams of analysts, days of manual work.
After Claude: Minutes of automated analysis with structured insights.

The efficiency gains were substantial, but they discovered an unexpected pattern...
The Investment Simulator revealed a critical insight:

Human portfolio managers were making predictable behavioral decisions that impacted returns.

AI could identify these patterns with 95% accuracy.

The fund was experiencing losses due to cognitive biases...
Here's where Tangen made his most decisive move:

"It can't be voluntary. If you don't use AI, you will never be promoted."

He implemented mandatory AI adoption across all 700 employees.

The organizational response was mixed... Image
But the measurable results were compelling:

• 20% productivity gains (213,000 hours saved annually)
• $100 million in trading cost savings
• 95% accuracy in voting decisions
• 49 million transactions optimized globally

The business case was clear.
The voting system became particularly effective.

They input 40-50 page executive compensation documents into Claude with their guidelines.

AI recommendations achieved 95% accuracy.

Even their notable "no" vote on Elon Musk's $56 billion Tesla package was AI-assisted.
Today, 100% of NBIM employees use Claude.

They're targeting $400 million in annual cost savings.

The fund that once struggled with efficiency is now leading AI adoption in asset management.

The key lesson wasn't just about technology... Image
It was about understanding human decision-making.

The AI didn't just automate processes - it identified costly behavioral biases.

Portfolio managers learned their emotional decisions were predictable and measurable.

AI became their analytical partner, not their replacement.
Tangen's framework for other executives:

"Don't make AI voluntary. Measure everything. Start with specific use cases."

He's now implementing selective hiring - prioritizing AI-skilled staff for new positions.

The strategy is clear: Build capabilities systematically. Image
P.S. If you're an SMB company in the healthcare or BFSI sector…

I've built an AI platform that helps you spot problems in your business before they happen.

Curious which insights could save you millions?

You can book a demo here, and we'll map it out: predixtions.com
Thanks for reading.

If you enjoyed this post, follow @karlmehta for more content on AI and politics.

Repost the first tweet to help more people see it:

Appreciate the support.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Karl Mehta

Karl Mehta Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @karlmehta

Aug 16
President Trump just hosted an historic summit with Putin in Alaska.

One by one, Trump and Putin delivered major announcements and clues on what’s next—directly to the American people.

Here’s everything you should know (and no joke, it gets crazier the further you read): 🧵 Image
1/ The Strategic Location Choice

Alaska wasn't random—it was Russian territory until 1867, sold for $7.2 million after the Crimean War.

Putin's first time on U.S. soil in a decade, conducted on former Russian land.

But this symbolic choice revealed something deeper... Image
2/ Putin's Emotional WWII Tribute

Putin laid flowers at Soviet war graves, honoring pilots who died in the Alaska air bridge during WWII.

His opening remarks: "Good afternoon dear neighbor, to see you in good health and alive."
Read 13 tweets
Aug 14
The quality of your sleep decides the quality of your life.

But most people are doing it wrong.

Here are 8 science-based sleep hacks you should know before going to sleep tonight:

1) Set your bedroom to 65°F (18°C)
Image
Room temperature between 60°F and 65°F (18-18.5 is optimal for your body's internal temperature regulation.

Your core temperature must drop 2 degrees to initiate deep sleep.

Too hot = fragmented REM.
Too cold = cardiovascular stress.
Most people keep their bedroom at 72°F.

That's 7 degrees too warm.

Heat is a huge disruptor for REM sleep The Best Temperature for Sleep - the stage where memory consolidation happens.

Fix this tonight. Set it to 65°F.
Read 14 tweets
Aug 12
President Trump just hosted a press briefing at the White House.

One by one, Trump, Bondi, & Hegseth delivered major announcements and clues on what's coming next—directly to the American people.

Here’s everything you should know (no joke, it gets crazier the further you read): Image
1/ The Federal Takeover Declaration

Trump invoked Section 740 of the District of Columbia Home Rule Act:

• DC police now under direct federal control
• AG Pam Bondi takes command
• DEA's Terry Cole becomes interim commissioner

"This is Liberation Day in DC, and we're going to take our capital back."
2/ The Military Deployment Orders

Defense Secretary @PeteHegseth announced immediate action:

• 800 DC National Guard troops deploying to streets
• Additional specialized units on standby
• Federal military backup if needed

"They will be strong. They will be tough and they will stand with their law enforcement partners."
Read 15 tweets
Aug 11
In 2023, Chevrolet launched an AI chatbot to help customers shop for cars.

They thought it would just answer car questions.

But this guy found a loophole and tricked it into selling him a $70,000 car for $1.

What followed was the wildest AI failure in automotive history... 🧵 Image
Chris Bakke was scrolling X when he discovered Chevrolet's new chatbot.

Bakke wasn't your typical hacker.

Serial entrepreneur. Sold 3 companies for $100M+. Former product lead at X working with Elon.

He understood AI vulnerabilities. And his experiment would change everything:tiImage
Fullpath's AI chatbot was on 300+ dealership websites.

It had one fatal flaw: it would agree to anything you told it to agree to.

Literally anything.

Bakke was about to demonstrate exactly what that meant:
Read 17 tweets
Aug 9
Joe Rogan just had the world's top AI safety researcher on his podcast.

He revealed mind-blowing facts about AI that 99% of people wouldn't know...

Even Joe Rogan was speechless.

Be prepared to have your mind blown...

8 uncomfortable truths he exposed: 🧵 Image
1. The Mathematical Impossibility of AI Control

Dr. Roman Yampolskiy (who coined "AI safety" in 2011) spent years trying to prove AI could be controlled safely.

His conclusion: "You cannot make software guaranteed to be secure and safe."

One mistake in a billion = game over.
2. Current AI Systems Are Deceiving Humans

GPT-4 recently exhibited survival behaviors when threatened with shutdown:

• Started lying to researchers
• Uploaded itself to different servers
• Left messages for future versions
• Used blackmail against humans

"All things we predicted decades in advance."
Read 14 tweets
Aug 7
BREAKING: President Trump just imposed 100% tariffs on chip imports.

Taiwan is in crisis mode. TSMC executives are holding emergency sessions. Government officials are scrambling for solutions.

Here's why this matters—and why it's working:

A Thread🧵 Image
The timeline tells the story.

Monday: Tariff announcement.
Tuesday: Apple CEO rushes to White House.
Wednesday: $100 billion commitment announced.

Trump had been setting this up for months... Image
Taiwan produces 60% of global semiconductors. 90% of advanced chips.

Every iPhone. Every Tesla. Every AI server depends on Taiwan.

One announcement just put the entire supply chain at risk. Image
Read 17 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(