Tensorify Documentation
Tensorify is a visual workflow automation platform for developers. Design backend automation on a canvas, run it locally or in the cloud, and debug every node with full visibility.
New to Tensorify? Follow the Quick Start Guide to build and run your first workflow in under 5 minutes.
You build workflows visually in the canvas editor at app.tensorify.io. Each workflow is a graph of nodes connected by edges. Tensorify compiles that graph into Python and executes it — either on your own infrastructure via the CLI, or in the Tensorify cloud.
- Design workflows visually using 30+ built-in plugin nodes
- Test from the canvas or locally with
tensorify runner start - Deploy via webhook, HTTP endpoint, or MCP tool with one click
- Monitor every run through the Jobs dashboard
- AI-native — use the MCP Platform Server to let AI agents (Cursor, Claude Desktop) manage your workspace, or expose workflows as callable MCP tools
| Section | What it covers |
|---|---|
| Quick Start | Build and run your first workflow in 5 minutes |
| Installation | CLI install (optional for self-hosted), Python setup, API keys |
| Core Concepts | Workflows, nodes, edges, execution modes, runners |
| CLI Reference | Full reference for init, runner, login, export, runtime, self-update |
| Running Workflows | The local dev loop and production execution |
| Deploying Workflows | Execution modes, webhook triggers, API endpoints |
| Plugins | All 30+ built-in plugins with reference pages |
| Guides | Step-by-step walkthroughs for real-world automation |
| Settings & API Keys | API keys, team members, workspace settings |
| Environment Variables | Team env vars and how workflows access them |
| Expressions | Template syntax, variable access, operators, and helper functions |
Tensorify generates readable Python that runs on your own infrastructure. There is no black-box cloud execution unless you choose it. You can inspect, export, and modify the generated code at any time.
No. Most workflows are built entirely with visual nodes — no code required. The Code node is available for custom logic when you need it.
Yes. Run tensorify runner start (or tensorify runner install for a system service) on any server with Python 3.9+ installed, then deploy workflows from the web UI. The runner maintains an outbound WebSocket to Tensorify — no inbound ports to expose, no firewall rules needed.
In CLI execution mode, payload data never leaves your infrastructure. Tensorify receives only the execution signal and run metadata for the visual debugger. Raw request bodies are processed locally.