● Covered by 1 source · 1 reportMedium impact1 neutral
Kairos: Open-Source Risk and Fraud Decisioning Platform with AI Agent Tuning
Key points
- Kairos is an open-source risk and fraud decisioning platform.
- It offers real-time decisions for sign-ups, logins, orders, and payments.
- AI coding agents (e.g., Claude Code, Cursor) can tune workflows and models.
- Includes 'Grayson' decision model for unstructured data with few labels.
Introduction to Kairos
Kairos is a newly released open-source platform designed for real-time risk and fraud decisioning. It processes events such as sign-ups, logins, orders, payments, and payouts, providing immediate decisions (approve, step up, review, or block) along with a risk score and reasons. The platform records each decision and the underlying signals, including subject history, shared devices, cards, and session data.
Key Differentiators and Features
Unlike commercial alternatives like Sardine, Sift, Riskified, and SEON, Kairos emphasizes tuning by coding agents and decision models that can detect fraud with limited initial labels. Its features include customizable workflows, queues for case review, and an MCP server for agent integration. The platform supports experiments where agents propose model variants, which Kairos then trains and tests on historical data.
The platform includes 'Grayson', a decision model that interprets requests in plain language and works effectively with few labels and unstructured data. Kairos also supports logistic regression and gradient-boosted trees (XGBoost-compatible) for modeling, which can be trained within the platform or uploaded.
Agent-Driven Workflow and Model Optimization
A core aspect of Kairos is its integration with AI coding agents (e.g., Claude Code, Cursor, Codex) via MCP. These agents can act as fraud ML engineers, defining workflows, specifying decision model queries, and identifying fields for scoring. Agents can also design experiments by proposing different model variants, such as standalone decision models, logistic regression, gradient-boosted trees, or stacked models.
Kairos automates the training and testing of these variants on historical data, evaluating them based on metrics like ROC-AUC, PR-AUC, gain over baseline, precision, and recall. The agent then selects the best-performing variant, explains its choice, and deploys it as a new workflow version, returning to re-evaluate as new labels become available.
Technical Access and Self-Hosting
Kairos provides SDKs for TypeScript (@finic-ai/kairos) and Python (finic-kairos), a command-line interface (CLI), and an HTTP API for integration. Users can self-host the platform using Docker, offering flexibility and control over their fraud detection infrastructure.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
The daily brief
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Today's brief
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's brief
New every morning, and the back catalogue is archived by date.
Primary sources
GitHub cmc_internal/api GitHub github/collect GitHub finic-ai/kairos GitHub _private/browser GitHub get-started/accessibility GitHub en/copilotReporting from
Kairos, an open-source risk and fraud decisioning platform, has been released as an alternative to existing commercial solutions. It features real-time decision-making, customizable workflows, and integrates AI coding agents for automated tuning and model experimentation. The platform aims to provide fraud detection capabilities with fewer initial labels by using decision models and agent-driven workflow optimization.