Guys, I want to continue organizing calls and webinars on topics around AI and modern technologies.
Lately, I've been experimenting a lot with self-improving agents. The first step in this direction was my anti-spam (I have a large community that is attacked by smart bots). For years, I developed it based on my ideas, but the approach of "letting the agent try and improve itself" easily outperformed all previous ones. In two days, the agent conducted 84 experiments, and the new model lets through half as much spam.
What we'll discuss:
- What RSI (recursive self-improvement) and autoresearch are, and where the catch is
- My experiment: how the cycle is structured, charts, what worked and what didn't
- How I apply the same ideas to my agents: memory, rules from mistakes, skills that are added after each run
- How to start such a cycle yourself
📆 Tuesday, November 10, 11:30am ET / 19:30 MSK
📍 Registration on Luma:
https://luma.com/sslznifi
#rsi@rvnikita_blog #agents@rvnikita_blog #claudecode@rvnikita_blog #ai@rvnikita_blog
