Show HN: Kairos – An Open-Source Risk and Fraud Decisioning Platform

A new open-source platform named Kairos has been launched to provide developers and businesses with an alternative to traditional real-time fraud decisioning tools such as Sift and Sardine. Developed to tackle two major hurdles in modern fraud detection—namely the scarcity of labeled training data and the rapid evolution of fraudster tactics—Kairos aims to modernize how companies protect their digital assets.

Traditional fraud detection systems heavily rely on machine learning models like XGBoost, which require extensive historical labeled data to function effectively. Many growing businesses lack this data. Kairos addresses this bottleneck by integrating ‘System One’ decision models alongside conventional machine learning approaches. This allows engineering teams to incorporate straightforward, natural language questions—such as checking whether an email address is temporary—directly into their decision logic without needing massive datasets.

to combat fraudsters who constantly modify their behavior to evade static rules, Kairos introduces a Model Context Protocol server. This feature enables AI agents like Claude Code or Cursor to autonomously define decision workflows, train models on fresh company data, and select the highest-performing variants. The platform is available for self-hosting via Docker or through a fully managed cloud service.

Sources: Hacker News

Aggregated summary. Original: Hacker News