The rise of Cursor: The $300M ARR AI tool that engineers can't stop using

Key takeaways

1. Cursor started out building tools for mechanical engineering. The team pivoted to programming after seeing a larger opportunity that also matched what they knew.
2. Truell says custom models built for specific tasks, such as advanced autocomplete, work alongside foundation models like GPT and give Cursor an edge, even though the company did not start by building models.
3. Heavy internal use of the product (dogfooding) and fast iteration helped Cursor meet real user needs.
4. Truell argues that AI moats look more like consumer moats than enterprise lock-in: you keep winning by improving the product and keeping users happy.
5. Cursor hires for intellectual curiosity, willingness to experiment and honesty, which Truell sees as the traits that build a resilient team.