Early Access Live Now

Causal Agent Platform

A reusable, production-grade AI agent architecture for rigorous causal inference and experimentation.

The Causal Agent Platform is now live for early access testing. Try out the interactive developer portal to run automated causal analyses on your own datasets.

Platform Architecture

Agent Workflow

An iterative tool-calling protocol powered by structured layers:

  • Knowledge Retrieval
  • Reasoning Traces
  • Hybrid Orchestration
  • Validation Checks

Advanced Estimators

Unified API for industry-standard causal inference methods:

  • Propensity Score Matching
  • Double Machine Learning
  • Difference-in-differences
  • Synthetic Control & BSTS

Production Scaffolding

Built to scale across chat, UI, and internal services:

  • API-First (FastAPI)
  • Background Jobs Queue
  • Typed Artifacts & Caching
  • MCP Surface Integration

End-to-End Causal Workflow

The platform explicitly guides the data science lifecycle, guaranteeing methodological rigor from raw data to final estimates.

Phase 1

Inspection & EDA

Automated profiling, missingness checks, outlier detection, and chart suggestions.

Phase 2

Causal Design

Observational and quasi-experimental workflows. Interactive DAG formulation for cross-sectional data.

Phase 3

Rigorous Analysis

Assumption verification, robustness checks, point estimates, and Heterogeneous Treatment Effects (HTE).

Phase 4

Artifacts & Reporting

Executable Python code generation and comprehensive final HTML summary reports.