Finic launches Kairos, open-source risk and fraud decisioning platform

Show HN: Kairos – Open‑source risk and fraud decisioning platform

Ayan from Finic announced the launch of Kairos, an open‑source alternative to commercial real‑time fraud decisioning platforms such as Sift and Sardine. Kairos is designed to address two key challenges that traditional fraud tools face.

First, many businesses lack the large, high‑quality labeled data sets required by machine‑learning models like XGBoost. Kairos mitigates this by incorporating “System One” decision models alongside conventional models such as XGBoost and logistic regression. This approach lets users embed plain‑language questions—e.g., “Is this a throwaway email”—into the decision workflow, reducing dependence on extensive labeled data and better capturing unstructured fraud signals that XGBoost struggles with.

Second, fraudsters continually evolve tactics, rendering static detection rules and models obsolete within weeks. To keep pace, Kairos introduces an MCP server that allows an agent—such as Claude Code, Cursor, or Codex—to define a decision workflow, train models on an organization’s own data, and select the variant that performs best. This reduces the need for a large data‑science team to constantly retrain models and update rules.

Kairos can be self‑hosted with Docker or accessed via the hosted service at https://kairos.finic.ai. The announcement received 37 points and four comments on Hacker News.