NodeFox
How it works

Build the workflow. Run the work.
Every step in between, visible.

Define the process using the visual graph. Connect the inputs, business rules and systems, then run a case and inspect the result. Improve the same workflow as your process changes.

From design to execution

From a business process to a running workflow.

  1. 01

    Define the work

    Choose the trigger, inputs, outcome and owner. Define the steps using the visual graph.

  2. 02

    Connect and control

    Add models, files, APIs and code. Map data between steps and wire in the conditions for actions.

  3. 03

    Execute and review

    Run in the browser or headlessly. Pause for the operator at human Decision steps.

  4. 04

    Inspect and extend

    Check the recorded outputs, measure the process and reuse tested networks in the next workflow.

The executable platform

Design the graph. Follow the work.

The network defines how information moves, which checks run and when an action is allowed. Watch the actual editor and execution experience, or explore the illustrative stages below.

Illustrative NodeFox interface and synthetic data. This local walkthrough explains the graph; it does not call AI, run the NodeFox engine or update external systems.

Reference
CASE-104
Amount (sample USD)
1200

Reader → model interpretation → Data mapping and Code check → human Decision → activated Writer.

Ready to explore

Inspect the input, then run the checks. The Writer remains behind the Decision step. A required reference must be present before review.

Underneath

The pieces that keep it explicit.

The saved graph is what executes. Dependencies and conditions define how work proceeds. Model output can vary and affect a branch; the orchestration rules remain explicit.

Ten node types

Reader, Writer, Conversation, Decision, Data, Code, Buffer, Wait, Global and Network: each does one kind of job.

Data edges and activation edges

Data edges carry values between steps; activation edges release a step. Flag a write as Activated and data alone can’t run it — only the approval branch can.

Typed JSON

Schemas describe the expected input and output shape. Configure validation and failure routes before allowing model-derived values to drive an action.

Human decisions

A Decision step can ask a person and hold the run, shown as “awaiting input”, until they answer.

Wait and Buffer

Parallel branches converge on purpose; state the run has gathered stays available to later steps.

Headless execution

Export compatible workflows to supported Linux or macOS runners. Schedules, folder watches and webhooks initiate runs; MCP-in-workflow and OAuth connections require a different integration path headlessly.
After the run

Inspect the recorded run, step by step.

When a result looks wrong, the run points at the step that produced it, with its inputs and output, instead of one oversized prompt.

Illustrative recorded-run view. Playback inspects saved output; it does not repeat model inference.

Map your workflow onto these steps.

Bring one recurring process. Discuss its inputs, checks, human decisions and system access—and whether a paid pilot or direct deployment is a practical next step.