Solo founder · Infrastructure & Architecture Lead · 2025–Present

Debutedge

A JSON-driven workflow runtime with a dual-plane architecture — Cloudflare edge and self-hosted Kubernetes — built on the principle of equal, production-grade paths rather than a primary and a fallback.

Debutedge is a JSON-driven workflow runtime built around a single architectural conviction: an edge deployment and an on-premises deployment should be equal, production-grade peers — not a primary plane and a fallback. I call this philosophy "equal paths."

The Dual-Plane Architecture

The edge plane runs on Cloudflare Workers — globally distributed, serverless, and cost-efficient for bursty or geographically dispersed traffic. The on-premises plane runs on a self-hosted stack built on Talos Linux, Knative, and Cilium — giving full control over data residency, networking, and cost for steady-state workloads.

Both planes implement the same JSON-driven workflow contracts. A workflow definition that runs on the Cloudflare plane runs identically on the on-prem plane, with the same state model and the same observable behaviour. Neither plane is a degraded mode of the other.

Validating Equal Paths

Before treating the two planes as truly interchangeable, I built a smoke test suite that exercises the full workflow surface against both deployments. The most recent run hit 40 out of 40 tests passing across both the LAN-hosted plane and the Cloudflare plane, including state parity checks that confirm a workflow's state transitions are observably identical regardless of which plane executed it.

A Deliberate Cost Strategy

The Cloudflare side of Debutedge isn't there because self-hosting is hard — it's deliberate cost arbitrage. Cloudflare's pricing model rewards spiky, globally distributed traffic, while the self-hosted plane stays warm and absorbs steady, predictable load efficiently. Running both, kept genuinely interchangeable, lets workloads land wherever they're cheapest without sacrificing capability.

AI-Native From the Ground Up

Debutedge is built with AI-driven operations in mind — workflows that agents can compose, inspect, and modify through well-defined interfaces rather than bespoke scripts. That same philosophy is what drew me to EmDash for this very site: a CMS that ships with a built-in MCP server and CLI is a natural sibling to a workflow runtime designed the same way.

Status

Debutedge is in active development. The architecture has been validated end-to-end across both planes; go-to-market and commercial positioning — including a written AI Native Enterprise whitepaper aimed at mid-sized companies — are the current focus.