Mastering NeuronFlow
The ultimate guide to building, automating, and scaling your processes with private AI workflows.
Overview
NeuronFlow is an AI-native automation platform that allows you to build powerful workflows using a visual editor. Whether you need to automate research, sync data between tools, or create intelligent AI agents, NeuronFlow provides the foundations for reliable execution.
Intent-Driven
Describe what you want to achieve, and our AI assistants will help you construct the logic, select nodes, and configure parameters in real-time.
Private & Secure
We prioritize your data privacy. Workflows run in an isolated environment, and your sensitive credentials are never exposed.
First Workflow
Build your first automated research agent in just a few steps.
Create a New Flow
From the Dashboard, click Create Workflow. Give it a name and click save.
Add a Trigger
Drag a Trigger node onto the canvas. Set it to 'Manual' so you can run it with a button click.
Connect an AI Agent
Find the AI Agent node in the library. Drag it and
connect the Trigger to it. In the prompt, type:
Write a haiku about {{input}}.
Press Run
Click the Run button. Enter a topic (like "Coffee") when prompted and watch the AI work!
Core Concepts
Understanding these four pillars will help you master any workflow.
Triggers
The "When" of your flow. A flow can start on a schedule, when an email arrives, or via a manual button click.
Nodes
The "What" of your flow. Each node performs a specific task, like searching the web, sending an email, or transforming data.
Edges (Connections)
The "Flow" of data. Edges define the path data takes from one node to the next.
Executions
A historical record of a flow running. You can view the logs for every execution to see exact results and errors.
Dashboard Guide
The Dashboard is your command center. It provides at-a-glance monitoring of all your automation activity.
- Recent Executions: See the status (Success/Failed) of flows that ran recently.
- Active Workflows: Manage your library of saved flows. Toggle them on/off or enter the editor.
- Quick Actions: Shortcuts to create new flows or check your quotas.
- Notifications: Alerts for flows that require manual approval or those that encountered errors.
Workflow Editor
Our visual editor is designed for speed and clarity. Here's how to navigate it:
The Canvas
Drag nodes to move them. Use your mouse wheel to zoom. Hold spacebar to pan across large flows.
Properties Panel
Click any node to open its settings on the right. This is where you configure prompts, URLs, and logic.
Variables & Logic
Connect nodes together by passing data using {{curly brackets}}.
// Get the result from a previous node
Hello {{nodes.node_name.result}}
// Access specific parts of the data
The price is {{input.price}}You can find a node's name by clicking on it in the editor. Variables allow the AI to "read" information from earlier in the process.
Durable Memory
Normally, flows forget everything once they finish. Durable Memory allows a flow to remember information across different days or runs.
Use Cases
- Storing the "last processed date" for a sync task.
- Building a chatbot that remembers a user's name between sessions.
- Accumulating a list of leads over multiple weekly runs.
AI & Logic Nodes
AI Agent
The core of any automation. Processes text using advanced models like GPT-4o or Claude 3.5.
- System Prompt: Tell the AI how to behave (e.g., "Act as a copywriter").
- Input: Accepts text or variables from previous steps.
- Capabilities: Summarization, Analysis, Content Generation, and Tool Invocation.
Code Tool
Run custom JavaScript or PHP code for complex data transformations that AI might struggle with.
- JavaScript/PHP: Full language support within an isolated sandbox.
- Global `input`: Access any workflow data directly via code.
- Output: Returns a structured object back to the workflow.
Approval
A "Human Gate" that pauses execution. Perfect for ensuring quality before sending emails or spending budget.
- Notification: Receive an alert in the dashboard when a flow is waiting.
- Review: Inspect the proposed data before clicking Approve or Reject.
Loop (Iterator)
Processes arrays of data one by one. Essential for bulk tasks like processing a list of leads.
- Input: Requires a list/array (e.g.,
{{nodes.search.results}}). - Done Branch: Triggers a final step once all items are finished.
Flow Control Nodes
Trigger
Determines when the flow starts. Supports Manual, Schedule (Crons), and Email monitors.
Webhook
Expose a unique URL that other apps (like Stripe or Shopify) can hit to start your workflow instantly.
Switch (If/Condition)
Routes data through different paths based on values. "If price > 100, go left; else go right."
Merge
Combines multiple parallel paths back into a single stream of data once they finish.
Data & Utility Nodes
Sync (Diff/Merge)
Analyzes two datasets and identifies what needs to be created, updated, or deleted. Perfect for database syncs.
HTTP Request
Connect to any external API. Supports all common methods (GET, POST, etc.) and custom headers.
Set Node
Manually define variables or reformat existing data to make it easier for later steps to read.
Sort
Organizes lists of data alphabetically, numerically, or by custom date fields.
Memory Node
Saves and retrieves data from the durable workflow database. Action types:
set, get, append.
Context Memory
Stores static facts or rules that are always injected into AI prompts for consistency.
Wait
Pauses the workflow for a set amount of time (seconds). Useful for rate-limiting calls to external APIs.
Output
The final node in a flow. Defines what text or data is returned as the final result of the execution.
Integrations
Connect NeuronFlow to the apps you use every day. We support secure OAuth connections for popular platforms.
Example Flows
SEO Content Researcher
Trigger (Schedule) → Search Tool → Loop → AI Agent (Analyze) → Google Sheet (Save).
Customer Support Assistant
Trigger (Webhook) → AI Agent (Categorize) → Switch → Slack (Notify Human).
Best Practices
Human Gating
Always use an Approval Node before sending automated communications to customers or publishing public content.
Keep Flows Small
Instead of one giant "Mega-Flow," break your tasks into 2-3 smaller workflows that talk to each other. This makes debugging much easier.