Runway adds natural-language Workflows to its Agent
Runway now lets creators build, run and edit its node-based Workflows through plain-language prompts inside Runway Agent, collapsing pipeline-building into conversation.
What launched
Runway introduced Workflows inside Runway Agent on July 24, letting users create, run and edit its node-based generative pipelines entirely through natural-language instructions. Instead of manually wiring nodes together, a creator can describe the desired multi-step output — chaining image, video and editing operations — and the Agent assembles or modifies the workflow graph accordingly. Runway framed the goal as producing higher-quality outputs at greater scale, and said the feature is available to try immediately.
How it fits Runway's stack
Runway launched its node-based Workflows editor in late 2025 as a way to build custom, repeatable creative pipelines, and rolled out Runway Agent earlier in 2026 as a conversational front end to its media models. Wednesday's update fuses the two: the structured power of a node graph with the accessibility of describing what you want in a sentence. That lowers the skill floor for building complex, reusable video and image pipelines while preserving the control that professional users rely on.
The broader pattern
The release fits a wider convergence between AIGC tools and agentic interfaces. Where creative software once demanded that users learn its node systems and parameter panels, agents increasingly translate intent directly into structured operations — a shift also visible in Midjourney's editing tools and in coding agents that generate build pipelines from prompts.
Why it matters
Node-based workflows are how serious creators achieve consistency and scale, but their complexity has kept them out of reach for many. By letting an agent author and revise those graphs from language, Runway is trying to make repeatable, production-grade generative pipelines accessible to a far broader base of creators — and to keep pace as rivals race to wrap generation capability in more capable agent layers.