This short post collects notes and links for two Y Combinator talks. Thumbnails below link to the videos; embedded players are included for easy viewing.
Videos
How to Build a Self-Improving Company with AI — Y Combinator
A few short takeaways and prompts for teams and builders:
- Focus on measurable feedback loops: define metrics, collect signals, and automate where the ROI is clear.
- Use AI to scale repeatable improvement tasks while keeping human oversight on strategy and safety.
- Start with small pilot systems that demonstrate measurable improvement before scaling.
How to Build an Internal AI Agent That Evolves Itself — YC Root Access
Short notes:
- Design agents with safety and guardrails; make evolution safe, observable, and reversible.
- Balance automation with human-in-the-loop checks for high-impact decisions.
- Emphasize evaluation metrics and continuous testing when enabling agent self-improvement.
Links
- Y Combinator — How to Build a Self-Improving Company with AI: https://www.youtube.com/watch?v=X_JsIHUfUjc
- YC Root Access — How to Build an Internal AI Agent That Evolves Itself: https://www.youtube.com/watch?v=DGD9b8K42lk
Screenshots: YouTube thumbnails (downloaded and included in this repo). If you prefer different screenshots or specific timestamps as images, say which and a screenshot can be extracted instead.