Introduction
AI infrastructure for transforming unstructured data into structured outputs
Welcome to bem. We provide AI infrastructure for transforming unstructured data—documents, images, audio, and video—into structured, actionable outputs and decisions.
What is bem?
bem decomposes complex cognitive tasks into atomic, optimizable functions rather than relying on brittle prompt chains or managing agents at scale. Our platform offers composable building blocks that you can arrange into workflows with defined paths, branching logic, and state management.
Core capabilities:
- Extract structured data from 20+ file formats including PDFs, images, spreadsheets, and audio
- Compose processing pipelines using primitives like Transform, Analyze, Route, Split, Join, and Enrich
- Deploy with built-in monitoring, evaluations, and webhook integrations
Built for production:
- Type-safe outputs with schema validation
- Automatic routing to human review when confidence is low
- SOC 2 Type 2 and HIPAA compliant
Getting Started
Quick Start
Build your first workflow in minutes with a step-by-step tutorial
System Overview
Understand bem's core primitives and how they connect
Workflows Explained
Learn how to orchestrate functions into processing pipelines
Dashboard Guide
Explore the bem UI for building and monitoring workflows
API Reference
This documentation includes a comprehensive list of all available API endpoints with their parameters and responses. Use the sidebar to navigate to specific endpoints for Functions, Workflows, Calls, and more.
For API authentication details, see Authentication.