LittleHorse
LittleHorse provides a Business-as-Code platform that orchestrates AI agents inside deterministic workflows, enabling enterprises to automate mission-critical processes. The platform uses the open-source LittleHorse Kernel workflow engine, with a Decision Workers pattern that gives agents read-only tool access to mitigate prompt injection attacks and hallucinations. It includes StreamSense for real-time Kafka event processing and Pony ID for human-in-the-loop governance.
- Enterprise Software
- AI Agents
- Artificial Intelligence
- Developer Tools
- Software Only
Funding
Founders
Founder details are not available yet.
Product
Problem
Business processes are fragmented across dozens of SaaS solutions, microservices, people, tribal knowledge, and legacy systems, which blocks agent deployment. Without an action layer, business process logic remains hidden in point-to-point connections between systems, queue consumers, emails, and spreadsheets, preventing reliable AI-driven automation.
Solution
LittleHorse provides a Business-as-Code platform that orchestrates agents inside deterministic workflows, allowing organizations to agentify demanding mission-critical processes. The platform uses the open-source LittleHorse Kernel workflow engine to handle execution, retries, and state management across microservices, while agents handle the fuzzy parts of processes. A Decision Workers pattern wraps agents as task workers with read-only tool access, so agents output structured decisions that the workflow executes, mitigating prompt injection attacks and hallucinations. The platform connects external systems via LittleHorse Connect, processes streams in real-time with StreamSense, and provides human-in-the-loop governance through Pony ID, all with end-to-end observability and audit trails.
Target Audience
Primary users are software engineering teams and enterprises across industries that need to orchestrate complex business processes involving microservices, AI agents, and external systems, particularly those with mission-critical or event-driven workflows requiring governance and auditability.
Features
- Deterministic workflow orchestration with the open-source LittleHorse Kernel, allowing engineers to define workflows in code using preferred languages (Java, Python, Go, etc.) without rewriting existing code
- Decision Workers pattern that separates agent intelligence from orchestration, giving agents read-only tool access to prevent unauthorized actions and prompt injection attacks
- ExternalEvent primitive for async callbacks, allowing workflows to pause and resume upon webhook or event arrival without polling or correlation logic
- Real-time event processing with StreamSense that integrates with Apache Kafka, including schema governance, anomaly detection, and event-triggered workflows
- Built-in end-to-end observability showing every workflow execution, task inputs/outputs, stacktraces, and data flow in a visual dashboard
- Human-in-the-loop governance via Pony ID with SSO, groups, permissions, and support for bring-your-own OIDC providers or built-in Keycloak
- Kafka Connect integration to trigger workflows from events, poll external systems, and sink data streams
- Declarative timeouts, retries, and exception handlers for robust failure management without custom infrastructure code