Skip to main content
This quickstart walks you end to end through the core BlobHub loop — create an Organization, create a Blob, run an Execution on its Draft Revision, then commit it — using the web application. Once you have done it once in the UI, the same flow is available through the REST API and the Python SDK.
1

Sign in to BlobHub

Open blobhub.io and sign in. You can use Google sign-in or continue anonymously if you just want to poke around. Your personal profile is created automatically on first sign-in.
2

Create an Organization

Every Blob lives inside an Organization. From your profile, create a new Organization — give it a short, URL-friendly alias (for example acme). The alias becomes part of every Blob URL inside it.
You can create multiple Organizations. Use them to separate personal experiments from team work, or to scope access for different collaborators.
3

Create your first Blob

Inside your Organization, create a new Blob and choose its Blob Type. For this quickstart, pick Workflow — it gives you something you can immediately execute. Give the Blob an alias (for example hello-workflow).See Blob Types for the full list of available types.
4

Edit the Draft Revision

Every Blob starts with a Draft Revision, which is also its default — the one every other surface resolves to until you say otherwise. Open the visual editor and drop in a couple of components — for example a Flow Start, a Code block that returns a simple value, and a Flow End. Connect them in order.The Draft stays mutable, and it is the only Revision that runs — feel free to iterate.
5

Run your first Execution

Open the Workflow Blob’s playground, create a Session, and start an Execution — this runs against the Draft, since a committed Revision cannot. You should see Execution Events stream in real time as components run.
If the Execution completes successfully, you have just exercised the entire BlobHub platform stack — storage, versioning, execution, and real-time events.
6

Commit the Revision

When you are happy with the Draft, commit it. Committing freezes the Revision’s content and its runtime alike — the Session you just used is closed along with it. Committing does not create a new Draft; run New when you are ready to keep editing.See Concepts → Revisions for the full lifecycle, including how to undo a commit or start a fresh Draft.
7

(Optional) Call it from code

You can drive the same loop programmatically. Install the Python SDK:
Then use it (or the raw REST API) to create Sessions and Executions against your Blob’s Draft from your own application or notebook — or read default to follow whichever Revision is current.

​
What You Just Built

This is the core loop for every Workflow Blob. Scheduler Blobs add time-based triggering on top of it, and ONNX Blobs swap the executable content for versioned model files — but the Organization → Blob → Revision shape stays the same everywhere.

​
Next Steps

Concepts

Get the full mental model of Organizations, Blobs, Revisions, and related primitives.

Blob Types

Explore Workflow, Scheduler, and ONNX Blob Types in depth.

REST API Reference

Drive the same loop programmatically from any language.

Python SDK

Integrate BlobHub into Python applications and notebooks.