Learnings from Cursor - 6 months later

In July 2024, I wrote my first post about Cursor. At the time, there were very few articles online about the company, and they had managed to stay almost entirely under the radar.

6 months later, Cursor is one of the hottest companies in Silicon Valley and recently announced a competitive $105M raise at a $2.5B valuation led by Thrive & a16z. They also shared the company is at $100M ARR.

There’s no other product (maybe besides ChatGPT?) that I’ve seen spread so quickly via word-of-mouth. Almost any founder I talk to wants to pay for it for their entire engineering team - as Cursor can help engineers code just a bit faster and solve problems they may not have experience in. I have also heard people say that they would pay a lot *more* for Cursor than they are currently paying. Developers are so valuable, that a slight increase in their productivity and happiness is usually worth paying for.

Cursor does not do any PR or marketing. I believe their team is small. Maybe less than 20 people?

Below are a few lessons I’ve distilled from this incredible company:

1. Quick product iteration

   1. Some developers I talked to tried Cursor and then churned off. However, when they go back and try it again a few months later - they often like it and stick with it. The Cursor / Anysphere team has been focused on continuously iterating on the product and adding incremental value each day. They recently launched their first agentic product as well, which developers enjoy thus far. The Cursor team has created a product that people love, which is easier said than done.

2. Bottoms-up sales motion

   1. Cursor is the perfect example of a PLG sales motion. Developers can sign up and easily use the product as a plug-in to GitHub, which they’re usually already using. They don’t need to go through extensive IT or security reviews, that I’m aware of. Then, when developers love it they tell their friends about it - and the product starts going viral.

3. Assistant vs. E2E (End-to-End)

   1. Cursor enables developers to work more quickly and is a smart assistant to help them code faster. Competitors like Cognition pitch themselves as a ‘replacement’ for entry-level engineers, which is not a message that resonates as well with developers. No one wants to be replaced. Cognition is also $400 a month, which is significantly more expensive.

Overall, the AI revolution is just getting started, and I imagine going forward there will be versions of Cursor for most roles where there is a scarcity of talent and resources. Tools that can help expensive talent become 10x or 100x more productive will almost always be worth paying for.