DagFlux Documentation
DagFlux is a desktop application for designing, automating, and running data workflows.
You assemble a workflow as a graph of typed nodes and let the built-in AI assistant
translate plain-language descriptions into the exact operations needed to move and reshape your data.
How DagFlux works in one minute
You start with a blank canvas and drop nodes onto it. Each node has a job:
- Start node decides when the workflow runs (manually, on a schedule, when a file changes, or when an HTTP webhook is called).
- Data Source node points at one input — a database table, a cloud bucket file, a local spreadsheet, a REST API, etc.
- Transformer node modifies data from a single upstream source.
- Join node merges data from two or more upstream sources.
- Output node writes results to a file, a database, a cloud bucket, a vector store, or an AI analysis report.
When you describe what you want a node to do in plain language, DagFlux asks clarifying questions if needed,
then generates the precise operation that will run at execution time. You see and approve that operation
before the workflow is saved.
When the workflow runs, DagFlux executes nodes in order, logs every step, and records the result so you can
inspect what happened.
Your credentials and AI provider keys stay encrypted on your machine and are never visible inside
generated workflows or sent anywhere except to the service the connection actually targets.
Documentation map
| Section |
What it covers |
| Architecture |
How DagFlux is structured, what runs locally, how runs are recorded |
| Plugins |
How DagFlux is extended with connectors and AI skills |
| Connections |
Saved data sources, credential storage, OAuth |
| AI providers |
Which AI providers DagFlux supports and how it uses them |
| Nodes |
What each node type does and how it behaves at run time |
| Chat prompts |
How the AI assistant turns descriptions into operations |
| Triggers |
Scheduled, file-based, and webhook automation |
What DagFlux can do
In one workflow you can:
- Pull data from any supported database (relational, document, or vector), cloud storage bucket, local file, spreadsheet service, or REST API.
- Reshape it with AI-generated SQL, NoSQL operations, or custom code, either creating a new dataset or modifying the source in place.
- Combine data from multiple sources with joins — including hybrid joins that mix relational, document, and vector sources, and joins that perform similarity search per row.
- Write the result to a local file (in many formats), back into a database, into a cloud bucket, or into a vector index (with embeddings generated automatically).
- Run the whole workflow on demand, on a schedule, when a file changes, or in response to an HTTP webhook.
- Automatically generate an AI analysis report on the final output.
DagFlux ships with a curated set of integrations and can install more on demand from the plugin catalog.
