Predictive Risk

Predictive Risk Analytics

ML-powered risk predictions with factor analysis, confidence scoring, and proactive mitigation for your AI agent fleet.

Start Free Read the Docs

Capabilities

Predict before it happens

ML Model Training

Train gradient boost, neural net, ensemble, or rule-based models on your agent behavioral data.

Risk Predictions

Get probability-scored predictions for policy violations, score drops, data breaches, and compliance failures.

Factor Analysis

Understand which factors contribute most to each risk prediction with detailed factor breakdowns.

Impact Assessment

Categorize risks from negligible to catastrophic with automated impact scoring based on historical data.

Trend Analysis

Track risk trends over time. Identify improving or deteriorating risk profiles across your agent fleet.

Proactive Mitigation

Automatically trigger alerts, sandbox agents, or activate kill switches when risk thresholds are breached.

Getting Started

How it works

Train Models

Select a model type and feature set. DRD trains on your historical agent data to build accurate risk predictors.

Generate Predictions

Models continuously evaluate agent behavior and generate risk predictions with probability and impact scores.

Act on Insights

Review predictions, understand contributing factors, and take proactive action before risks materialize.

Developer Integration

Predict programmatically

risk.ts

import { DRD } from '@drd/sdk';

const drd = new DRD({ token: 'drd_live_sk_...' });

// Train a risk model
const model = await drd.risk.trainModel({
  name: 'violation-predictor',
  modelType: 'gradient_boost',
  featureSet: { features: ['action_freq', 'score_trend'] },
});

// Get risk predictions
const predictions = await drd.risk.listPredictions({
  riskType: 'policy_violation',
});

for (const p of predictions) {
  console.log(`${p.riskType}: ${p.probability}% (${p.impact})`);
}

Predict risk before it strikes

Train ML models on your data. Get predictions before problems materialize. Act proactively.

Get started free View documentation

import { DRD } from '@drd/sdk'; const drd = new DRD({ token: 'drd_live_sk_...' }); // Train a risk model const model = await drd.risk.trainModel({ name: 'violation-predictor', modelType: 'gradient_boost', featureSet: { features: ['action_freq', 'score_trend'] }, }); // Get risk predictions const predictions = await drd.risk.listPredictions({ riskType: 'policy_violation', }); for (const p of predictions) { console.log(`${p.riskType}: ${p.probability}% (${p.impact})`); }