dlt and dltHub

note

dltHub offers two products: dlt (open source) and dltHub (commercial, license-gated). This page explains both products and how they relate.

• dlt — the open source ingestion library, Apache 2.0.
• dltHub — the agentic platform that deploys, monitors, and scales dlt pipelines, with a managed runtime, data quality, transformations, and AI tooling for coding agents. All dltHub components are license-gated.

The two products at a glance

dlt is free and open source under Apache 2.0, dltHub is a paid product.

Capability	dlt	dltHub
Build ingestion pipelines using dlt verified sources (except premium sources)	✅	✅
Build pipelines with the dltHub AI Workbench	—	✅
Data quality metrics & checks	—	✅
Build transformation pipelines (dltHub/dbt)	—	✅
Managed runtime: deploy, run/schedule pipelines, serve data apps, monitor jobs	—	✅
Premium sources (MS SQL) and destinations (Iceberg, Delta, Snowflake + Iceberg/Open Catalog)	—	✅

dlt (OSS)

dlt is the open source ingestion library, distributed under Apache 2.0.

Component	Description	How to access it	Get started
dlt library	The Python pipeline engine: extract, normalize, load. Includes schema inference and evolution, incremental loading, write dispositions, pipeline state, and the dlt pipeline … CLI for inspection.	dlt — Apache 2.0, on PyPI	Intro · Tutorial · pip install dlt
Core sources	Three flexible, generic sources shipped with the library that cover most ingestion scenarios out of the box: rest_api (any REST API via declarative config for endpoints, pagination, and auth) sql_database (30+ SQL databases via SQLAlchemy / PyArrow / pandas / ConnectorX with table reflection) filesystem (local and cloud storage — S3, GCS, Azure Blob, Google Drive, SFTP — with native CSV / Parquet / JSONL support)	Distributed with dlt library	REST API · SQL Database · Filesystem & Cloud Storages
Verified sources	A curated set of dltHub-maintained connectors (for example, Kafka, MongoDB, Postgres CDC, Stripe, Hubspot, …) pulled into your project with dlt init <source> <destination> .	dlt-hub/verified-sources	Verified sources docs · dlt init -l to list available sources

dlt is a good fit if: You want a lightweight, code-first ingestion library, are comfortable managing orchestration, scheduling, and operations yourself, or you need to deploy on-prem, on a VPS, or in any environment where managed cloud solution is not an option. dlt runs anywhere Python runs, with no platform dependency.

dltHub

dltHub is a managed cloud platform for running your dlt pipelines, transformations, and notebooks. You can work with dltHub in two complementary ways:

• Web UI at app.dlthub.com — sign up to deploy, schedule, monitor pipelines, manage profiles, browse datasets.
• Locally, from the CLI or Python — bootstrap a new workspace in one command:

  uvx dlthub-start@latest my-workspace

  This creates a runnable workspace with the AI Workbench, example pipelines, and the dlt[hub] extra installed. To add dltHub to an existing project instead, run:

  pip install "dlt[hub]"

  Either way, you get the dltHub workspace and dashboard, the AI development tooling (dlthub ai, MCP server, AI Workbench), per-source contexts, and the dlthub library that adds data quality, transformations, and premium sources/destinations.

Every component below is part of dltHub and requires a license. Most components are source-available under their own licenses; all are distributed through the dlthub PyPI package or the dltHub repositories.

Component	What it is	How to access it	Get started
dltHub Platform	The hosted Web UI and managed runtime at app.dlthub.com — deploy and schedule pipelines, monitor runs, manage workspaces and profiles, browse datasets, collaborate.	app.dlthub.com	Platform · Runtime overview · Runtime tutorial
AI Toolkits	The dltHub AI Workbench: a collection of toolkits made of skills, rules, workflows, and MCP wiring that drive agentic pipeline development inside Claude Code, Cursor, and Codex.	dlt-hub/dlthub-ai-workbench — source-available under its own license	Agent-native workflow walkthrough
dltHub Context	Per-source agent contexts (specs, endpoint documentation, prompts) that prime your coding assistant for thousands of APIs. Automatically used by AI Workbench	Browse and copy contexts at dlthub.com/context	Build a source with AI
dlthub library	Python package shipped via dlt[hub]. Adds the production capabilities: data quality, Python transformations (@dlt.hub.transformation) and dbt transformations, and premium sources/destinations such as Iceberg / DuckLake and MSSQL Change Tracking.	On PyPI	Installation

dltHub is a good fit if: You are running pipelines in production, want a coding agent to do the heavy lifting with tooling that supports the generation of production-grade code, need transformations or data quality checks, if you want managed infrastructure, or if you are working as a team.

tip

If you have a specific question, feature request, or unique use case, feel free to reach out.

Getting started

Bootstrap a new dltHub workspace in one command:

uvx dlthub-start@latest my-workspace