Hyphen is built on four layers, each responsible for a distinct concern. Together, they bridge rule-based automation with AI reasoning while keeping everything governed and auditable.
This page is for technical evaluators. For business outcomes, see Solutions. For deployment options, see Enterprise.
The core that manages workflow execution, agent reasoning loops, state transitions, and context accumulation across steps.
Specialized engines for operations that should never rely on LLM inference: matching, validation, human approval, and document handling.
Defense layers that run on every request: sanitizing inputs, redacting sensitive data, and preventing unauthorized agent behavior.
The distributed systems patterns that ensure workflows complete even when things go wrong: server failures, timeouts, and resource exhaustion.
The full component reference lives in the docs.
No proprietary runtime, no exotic dependencies. Hyphen runs on standard infrastructure you already operate, in your own cloud or on-prem.
Our team will take you through the architecture layer by layer, tailored to your use case.