Most AI workflows never get usedI’m building what makes them runnable, usable, and worth paying for

Founder of Edgaze, a marketplace and hosted runtime where AI workflows are built, published, and run per use. Everything here is a record of that work.

Arjun Kuttikkat

Building now

Inside Edgaze

A published workflow is one unit. It is built on a canvas, executed by a hosted runtime, sold per run on a marketplace, and callable from a backend or an agent at the same price.

Edgaze Workflow Studio with a simple three-block workflow on the canvas. Block library on the left, inspector on the right.
Workflow Studio. A template on the canvas: Input → LLM Chat → Output.
  1. 01Build

    Workflow Studio

    A node-based canvas with 18 block types, templates, Composer for AI-assisted edits, and a vault for provider keys.

  2. 02Execute

    Hosted runtime

    An execution engine and workers run the graph. Every run is version-pinned and persisted with a durable event log.

  3. 03Distribute

    Marketplace

    Publishing creates a product page with a price per run. Buyers pay from a wallet or bundle; creators keep 80% of margin.

  4. 04Run from anywhere

    API and MCP

    The same published workflow runs from a backend over REST, with SSE and signed webhooks, or from an agent over MCP.

Terminal

The site has another interface.

A functional shell over the portfolio. Browse projects and writing, inspect Edgaze, navigate a virtual filesystem, open pages, or subscribe without leaving the terminal.

visitor — -zsh — 80×24

arjunkuttikkat.com · terminal 1.1.0

Type 'help' to see available commands. Tab completes.

 

Try
Arjun Kuttikkat

About

Founder of Edgaze. Robotics and AI student in Dubai, from Kerala.

I grew up in a small town in Kerala and started building early. Most of it failed: five YouTube channels, no traction. At 18 I moved to Dubai, incorporated Edge Platforms, Inc., and started Edgaze. It is live and operating in production, and acquiring creators and buyers is most of what I do now.

Read more