Build LLM pipelines in VS Code, run them on a C++ engine

What it does

RocketRide is an open-source pipeline builder for AI and ML workloads. You assemble workflows in a visual canvas inside VS Code, save them as portable JSON *.pipe files, and execute them on a multithreaded C++ runtime that stays on your own hardware. It ships with 50+ nodes spanning 13 LLM providers, eight vector databases, OCR, NER, and chunking, plus Python and TypeScript SDKs to embed pipelines into existing applications.

The interesting bit

Instead of context-switching between an editor and a separate dashboard, RocketRide moves the visual builder directly into VS Code. The project also auto-detects coding agents like Claude and Cursor, letting you build or modify pipelines through natural language without leaving the IDE.

Key highlights

• All nodes are Python-extensible, so you can build and publish your own nodes in Python
• Built-in observability traces token usage, latency, and memory consumption per pipeline run
• Supports CrewAI and LangChain for multi-agent workflows with shared memory across runs
• MCP server support exposes pipelines as callable tools for other AI assistants
• Stays on your own hardware through local, Docker, or on-premise deployment options

Verdict

Worth evaluating if you want a visual AI pipeline builder that lives inside VS Code and runs on your own infrastructure. Skip it if you prefer writing pure code to dragging nodes, or if you are waiting for the fully managed cloud option that is still on the roadmap.