Stof Workflows

A singular & portable AI workflow artifact

AI workflows are currently a mess, being fragmented across:

• Langchain/LlamaIndex (Python libraries with tons of dependencies)

• N8N/Zapier (visual tools with vendor lock-in)

• Custom scripts that are scattered, fragile, and difficult to maintain

• YAML, TOML, JSON, etc., configs that need separate execution engines

No current solution offers a combination of portability + programmability + safety for fully describing and executing AI workflows.

Write Once, Run Anywhere

Stof documents are:

• Portable: Send the entire workflow artifact over an API or store it in a database

• Self-describing: The data, validation, and execution logic travel together

• Versionable: Track workflow evolution like code

• Inspectable: See exactly what a workflow does without executing it

• Cloud-agnostic: Run on AWS/GCP/Azure/on-prem

Not Another Library

Stof is a standard rather than a library. Instead of "yet another config format", Stof is a portable, self-contained workflow artifact that can run anywhere.

What really differentiates Stof in this landscape is that Stof workflows are transmissible artifacts. You can:

• Email a workflow

• Store workflows in Postgres

• Version workflows in Git like they're data (because they are)

• Compose workflows by importing them

• Execute untrusted workflows safely (sandboxing)