InitfinityTM

Visual workflow builder

Visual where it helps, code where it counts

Initfinity is a node-based workflow editor that runs right in your browser. Wire up HTTP calls, data, AI and logic, drop into JavaScript or Python when you need it, then host the flow on a schedule or a webhook.

Create your FlowRead the docs

No account needed to start. Client-side flows are free forever.

uptime-monitor.flow
yesnoScheduleevery 5 minutesHTTP FetchGET /healthConditionis the site down?AI Summarizewrite the alertSlackpost to #opsLogsite is fine
100+built-in nodes
No accountneeded to start
3interchangeable flow formats
1 clickto host on a schedule

Why Initfinity

Everything you need to automate, nothing to install

01 / Local-first

Runs in your browser

Flows execute on your device, so you can build and run them without an account. Your work stays in your browser unless you choose a cloud feature.

02 / AI

AI-assisted building

Describe what you want and Generate Flow drafts the graph. Refine it by hand on the canvas.

03 / Hosting

Host it when it matters

Put a flow on a schedule or behind a webhook URL and let it run without a browser tab.

04 / Together

Built to collaborate

Edit the same flow in a live room, or publish a read-only copy anyone can view and fork.

05 / Nodes

100+ built-in nodes

HTTP, JSON and CSV, data frames, charts, scripting, crypto, timers, even Bluetooth, serial and USB.

06 / Open formats

Flows you can read, diff and version

Convert losslessly between a readable text format, JSON and a compact binary format, so flows fit version control and AI tools.

FDLJSONFLOB

Use cases

One canvas, whatever you need to automate

Hear about downtime before your users do

Check a URL on a schedule and post to chat only when it fails, so the channel stays quiet when everything is fine.

Open this flow →
uptime-monitor.flow
yesScheduleevery 5 minutesHTTP FetchGET your siteConditionis the site down?Slacksite is down

Let AI read the inbox and flag what matters

Classify incoming messages with a model and alert your team only when one is negative. Bring your own key or use the built-in AI.

Open this flow →
ai-sentiment.flow
yesWebhooknew messageAI ClassifysentimentConditionis negativeSlackalert the team

Connect two services that do not know each other

Receive a webhook, reshape the payload with a few nodes and forward it wherever it needs to go.

Open this flow →
webhook-relay.flow
Webhookincoming POSTJSON Pathpick the fieldsDiscordforward it

For developers

Drop into code when the canvas is not enough

Mix visual nodes with real JavaScript, Python or WebAssembly. Heavy work runs in a pool of web workers, so the canvas stays responsive.

analyze.py · Code node
# Code node (Python), runs in your browser
import numpy as np

readings = np.array(inputs["in"]["values"], dtype=float)
spread = 2 * readings.std()

result = {"out": {
    "mean": float(readings.mean()),
    "outliers": int((abs(readings - readings.mean()) > spread).sum()),
}}

Dependencies: numpy

terminal · Alt+S
$ lsuptime-monitor.flow  health-check.fdl  notes.md$ flob pack uptime-monitor.flowPacked /uptime-monitor.flow → /uptime-monitor.flob (2.4 KB → 789 B, 68% smaller)$ load uptime-monitor.flowLoaded flow from /uptime-monitor.flow$ runStarting flow execution…$ hostOpening the host dialog…

JavaScript or Python

Write a Code node in either language. Python runs on Pyodide in the browser, and both can pull in packages, from PyPI for Python and npm for JavaScript.

WebAssembly

Load a .wasm module as a node, as a plain module, a WASI program or a component, and call it from your flow.

A terminal in the editor

Press Alt+S for a command-line shell over your files and flows. Run, host and stop flows, convert formats and script your workspace from the command line.

A real filesystem

Flows, scripts and data live in a browser-based filesystem, with version history built in.

Plugins

Plugins are plain JavaScript modules. Register terminal commands, inject UI and hook into the engine.

Agent-ready

Flows are readable text, and a WebMCP tool surface lets AI agents read and author them.

Read the scripting guide →

Open format

Your flow is also just text

FDL is a readable format that converts losslessly to and from the graph. Write it by hand, generate it with AI or keep it in git.

health-check.fdl
#fdl 1
flow "Health check"

# Each line is a node or a connection
tick(interval) interval=300000
ping(http-fetch) url="https://api.example.com/health" method=GET
log(logger) prefix="health"

tick.out -> ping.in
ping.out -> log.in
the same flow, as a graph
tickintervalpinghttp-fetchloglogger

Read the FDL guide →

How it works

From idea to running automation in three steps

  1. Build on the canvasDrag nodes onto the canvas and connect their ports. Each node does one job, like fetching a URL or filtering rows.
  2. Run and inspectConfigure nodes in the Inspector, run the flow and follow the data from node to node to see exactly what happened.
  3. Host or shareTurn on a schedule or webhook trigger to run it in the cloud, or publish a copy for others to view and fork.

Examples

Start from a flow someone already built

These are public flows. Open one to see how it is wired, then fork it into your own editor.

Website uptime monitor to Slack

Checks a URL every 5 minutes and posts to chat only when the site is down.

AI sentiment triage

Classifies incoming messages with AI and alerts chat only when one is negative.

Webhook relay and transform

Receives a webhook, reshapes the payload, and forwards it to Slack or Discord.

Daily AI news digest

Fetches a list each morning, has an AI condense it, and posts the digest to chat.

Price threshold alert

Checks a price every 15 minutes and alerts chat when it crosses your threshold.

Contact form to chat

Turns a form submission into a chat message and drops entries without a valid email.

Create your first Flow

No sign-up needed. Open the editor, drop in some nodes and run it.

Create your Flow