posts/
20 pages · Updated April 30, 2026
Pages
- posts/how-to-win-hackathons/index.html
- Building the Evaluation Engine
- The results [Link to The results](/content/posts/what-model-to-use#the-results/index.html)
- The Benchmark [Link to The Benchmark](/content/posts/online-mind2web-benchmark#the-benchmark/index.html)
- Why stealth needs its own benchmark
- Training beats prompting. Every time. [Link to Training beats prompting. Every time.](/content/posts/agent-freedom#training-beats-prompting-every-time/index.html)
- Why abstractions break learning
- How it works
- posts/web-scraping-guide-2026/index.html
- Cookie Syncing [Link to Cookie Syncing](/content/posts/web-agent-authentication#cookie-syncing/index.html)
- posts/index.html
- How we got here
- Kepler and Agent Score
- Our Mission: The Default Layer for Browser Automation
- The cat & mouse game is about to get harder
- The learning
- 100 hard browser tasks, one leaderboard
- The Problem
- Exploration vs exploitation
- Introducing BUX