How it works
Connect your repository.And start shipping.
Just attach your repo. We handle the path from CI and the build to a production-ready deploy, including your custom domain.
What dFlow is
The app you already have. In production.
Connect the product you already built. We handle the infrastructure so you can keep building.
Apps, databases, domains, logs, and deploys stay together.
We read the source, build it, and put the app in production. Cloudflare is the hostname and SSL path.
Your stack. Our servers, Azure, or a machine you attach. AWS, GCP, and DigitalOcean attach now.
Built with
The stack this repo runs on.
The control plane is Next.js and Payload. Mongo holds product state. Redis and BullMQ run jobs. Dokku hosts that app. Tailscale is how the control plane reaches worker nodes, not a customer VPN. This is our stack, not the one you deploy.
- AppNext.js + Payload
- StateMongoDB + Redis
- HostDokku + Tailscale
not your deploy
Railpack, Dockerfile, Static, or BuildPacks build the repo you connect. That is a different beat.
Control plane
Create an application. Pick an environment.
That is where the wizard starts. An application holds environments. An environment is bound to compute. Services (apps, databases, images) live under it.
You push
Git push
GitHub · GitLab · Bitbucket · Azure DevOps · Gitea
Docker image
Any registry
Archive upload
zip · tar · tar.gz
Object model
Environments
Staging and production. Same path.
Name the environment. It runs on one machine. You do not invent a second setup.
storefront
production
one worker · one bind
Compute binding
No vendor lock-in. Deploy stays the same.
Use our affordable compute, your Azure credits, or a machine you already have. AWS, GCP, and DigitalOcean attach now. Full setup on those clouds comes later.
bound to
dFlow compute
worker-prod
Connect now · full setup later
- AWS
- GCP
- DigitalOcean
Detected stacks
Bring the stack you already know.
We find it. Switch the builder only if you need to. No repo? Send a Docker image or an archive.
Stack
Use the stack you already know
Bring the app you already have. We run it in production.
Next.js
runtime · Railpack
runtimes
data
same detect
- Node
- Python
- Go
- PHP
- Java
- Ruby
- Rust
Four products
Apps, data, templates, and your editor.
Deploy and manage apps, databases, and templates in one place. Ask your AI agent to handle it.
- Workspace
- Application
- Environment
- app
git, image, or archive on the worker bound to that environment.
- Workspace
- Application
- Environment
- database
Six engines attach on a private connection. ClickHouse backups are not supported.
- Workspace
- Application
- Environment
- wired services
A catalog stack lands as services already wired, in start order.
- Editor
- scoped tools
- same objects
The editor inspects deploys, reads logs, and ships the next change. Credentials stay in your account.
Editor plane
Your editor ships the same path.
Ask Cursor or Claude to inspect the deploy, review the logs, and keep shipping. Credentials stay in your account.
- AWS
- GCP
- Azure
- Hetzner
- DigitalOcean
+ dFlow compute · VPS · on-prem
- 01Dashboardthe same UI
- 02Gitpush to your node
- 03EditorMCP over OAuth
on the node bound to the environment
- Antigravity
- Cursor
- Claude Code
- Codex
- GitHub Copilot
MCP over OAuth
- Postgres
- MySQL
- n8n
- WordPress
- Redis
+ any Docker image · templates
- create_applicationstorefront
- create_database_servicePostgres 16
- create_deploymentstorefront.up.dflow.sh
assembled
Build the product.We manage the infrastructure.
Same path from git, the dashboard, or your editor. On our servers, your cloud credits, or a machine you attach.