AI
AI Analysis
Live Data

Karpathy's 2-Hour LLM Workflow: Real-World Use Cases

Viral tweet of Karpathy's 2-hour LLM walkthrough analyzed: focus on practical workflows over benchmarks. Sentiment: 52.9% supportive, 41.2% confronting.

@RoundtableSpaceposted on X

ANDREJ KARPATHY JUST DROPPED A 2 HOUR WALKTHROUGH OF HOW HE ACTUALLY USES LLMS IN REAL LIFE. Not theory, not benchmarks, just the real workflow behind one of the sharpest minds in AI. https://t.co/ApFwvRrTL1

View original tweet on X →

Community Sentiment Analysis

Real-time analysis of public opinion and engagement

Sentiment Distribution

94% Engaged
53% Positive
41% Negative
Positive
53%
Negative
41%
Neutral
6%

Key Takeaways

What the community is saying — both sides

Supporting

1

Skill lives in the thinking process

Watching how Andrej thinks through a problem and iterates shows practical know-how you can't get from papers alone.

2

Transparency raises the bar for execution

Public walkthroughs put more pressure on developers to deliver tangible results now, not some distant quarter.

3

Real-world use beats benchmarks and papers

Viewers care more about how the tech is actually used than raw metrics or published numbers.

4

Rare access to masters' behind-the-scenes

Seeing an expert’s hands-on workflow reveals what truly works, not just academic theory.

5

Demand for concise takeaways

Replies asking for a summary reflect a clear appetite for distilled, actionable insights from long sessions.

6

Recency influences relevance

Questions like “was this recorded 2025?” show that people judge value based on how current the material is.

Opposing

1

old

repeated notes that it was “posted over a year ago” and shouldn’t be billed as “just dropped.”

2

misrepresenting the source

turning a general-audience “deep dive” into a sensational “HOW HE ACTUALLY USES LLMs” claim for engagement.

3

“Can we do something about it?”

and direct admonitions like “audit your posts before they ship” urging moderation or higher editorial standards.

4

“you sux man”

, “just dropped 🤣”) aimed at the poster rather than the content.

Top Reactions

Most popular replies, ranked by engagement

F

@fadeleusdev

Opposing

Bro this is from a year ago https://t.co/Djz2FlFQfF

6
0
789
C

@crashev

Opposing

Just dropped? Is 1 year ago JUST DROPPED for you ?

6
1
522
K

@kumarvinayak490

Opposing

@nikitabier This guy is using terms like “just dropped” for a video which was posted in 2025. Can we do something about it ?

5
0
369
M

@MonacoMacro

Supporting

@grok summarize this

1
2
168
T

@tinkermight

Supporting

this was recorded 2025?

1
1
79
A

@AstroFahim

Supporting

this is what separates knowing about ai from actually building with it. everyone can read papers. watching how andrej actually thinks through a problem and adjusts is the rare thing. that's where the skill lives.

0
1
340

This article was AI-generated from real-time signals discovered by PureFeed.

PureFeed scans X/Twitter 24/7 and turns the noise into actionable intelligence. Create your own signals and get a personalized feed of what actually matters.

Report an Issue

Found something wrong with this article? Let us know and we'll look into it.