Carlin Wiegner > AI > Cursor

Cursor

Why they can win?

• Move fast and experiment is a Facebook skill not a Microsoft skill. Cursor will get good at:
• Iteration speed
• Need for experimentation
• Unlikely Microsoft or anyone else can deploy this level team with this level incentives

• Developers have hard job and are very savy of using tools to assist.
• It isn't really just take an IDE and bolt on an AI model. There are hundreds of little features that are quite hard to make awesome. This is the iPod nuance.
• Individual buyers
• who has the lead can extend the lead

• math olympiad at 8:30
• why they'll beat microsoft? at 10:00
• which llm is better? at 37:00
• some dirty facts like models are focusing on benchmarks versus general intelligence
• benchmarks are in the training data - they even have the path data of the files
• minute 42 - vibe check
• 43:00 - it's hard to overstate how hard bugs are
• 1h 42m - the dangers of centralized computing
• 2h 06m - fields medal before AGI
• 2h 08m - Original AI law, Chinchilla paper
• 2h 26m - The best programmers

• Ignore the idea as much as possible that AI apocalypse is imminent.
• We are really like to have lots of models rather than one magic big model.
• AI and human computer interaction are different.
• HCI is still very promising area of thought.
• Pure AI is like Tesla FSD
• HCI is about knowing what human is trying to do and how to augment

• Series A on August 9, 2024 $60m at $340 million. Noah Brown, A16z, Jeff Dean, OpenAI, Patrick Collision, Thrive
• 10 employees as of October 9th, 2024

• Michael Truell
• MIT
• Graduated
• Sualeh Asif
• Personal home page
• Represented Pakinstan at International Math Olympiad
• Graduated MIT
• Arvid Lunnemark
• CTO
• International Math Olympiad Gold and Silver medal winner)
• Arman Sanger
• Personal home page
• Graduated MIT
• Seems very AI deep for oustide a frontier lab
• Clear performance engineering mindset

Created: October 9 2024.
Modified: October 9 2024.