The walkthrough on this page covers 6 steps: 1. Create a workspace account 2. Connect a source and a warehouse destination 3. Run the initial sync 4. Add SQL models from GitHub or dbt 5. Orchestrate and activate the results 6. Connect an agent through the MCP server.

What are the limitations of Weld?

Querying is blocked until an initial sync finishes processing full history, which can delay time to first insight on large sources. No pricing information is published on the homepage, so cost has to come from a sales conversation. The agent and API story assumes comfort with MCP configuration, API keys and warehouse schemas, so non-technical users will struggle.

Weld

Name: Weld
Author: Weld

Autonomous ETL pipelines that feed warehouses, BI tools and AI agents

Advanced API

Visit Weld Back to AI tools

Own Weld? Verify ownership to get a dofollow link

FeaturedHand-picked by aiwizard

AI Angels400+ customizable AI girlfriends for NSFW chat, voice calls, and roleplay

Candy AIExperience the future of companionship with AI friends who chat, role-play, and connect with you on a human level.

ourdream.aiA vast AI roleplay platform with millions of characters for chat, images, video and voice

Screenshot of Weld, Autonomous ETL pipelines that feed warehouses, BI tools and AI agents

What is Weld?

Weld is an autonomous ETL and data platform: 300+ connectors into BigQuery, Snowflake, Databricks and more, SQL and dbt transformations, orchestration, reverse ETL, and an MCP server that lets Claude or Cursor operate pipelines directly. Aimed at data and analytics teams that want one platform instead of several.

Weld is an autonomous data platform that moves data from applications, databases, files and event streams into a warehouse, then models and activates it. It bundles four jobs that teams usually stitch together from separate vendors: ingestion through incremental, full and CDC replication, SQL transformation with GitHub, dbt, Cube or SQLMesh models, orchestration with dependencies, schedules and retries, and reverse ETL that pushes warehouse results back into CRMs, ad platforms and internal tools. Each component can run alone or alongside an existing warehouse and modeling stack. The product targets data and analytics teams at companies that have outgrown hand-rolled Python scripts and cron jobs but do not want a heavyweight enterprise stack. Marketing and operations teams appear alongside data engineers in the published customer stories, which suggests the syncing and reverse ETL sides get real use outside pure engineering. A 4.8 rating on G2 with 109 reviews is the main third-party signal cited on the homepage. In practice a team connects a source such as HubSpot, Salesforce, Postgres or a Google Sheet, maps schemas to the destination warehouse, and schedules the sync. Transformations then live in version control: models are written as SQL and published from GitHub or synced with an existing dbt or Cube project. Orchestration runs the graph in order and retries failed steps, with tests and lineage attached to each model. The distinctive layer is the MCP server, which exposes models, syncs and pipeline runs as tools that Claude, Cursor, ChatGPT or Copilot can call, plus a REST API for embedding the same operations in an internal product. Weld sits in the modern data stack rather than replacing the warehouse. It competes with managed ELT services on the ingestion side and with orchestrators and reverse ETL vendors on the activation side, and its pitch is consolidation. Initial syncs process full history and querying is unavailable until they finish, so onboarding is not instant on large sources.

How do you use Weld?

1Create a workspace account
Sign up at the Weld workspace to get access to the console where sources, syncs and models are managed.
2Connect a source and a warehouse destination
Pick from the listed connectors, such as HubSpot, Postgres or Google Sheets, and point the sync at a warehouse like BigQuery or Snowflake.
3Run the initial sync
Start a full or incremental replication and monitor progress in the sync table. Querying stays unavailable until historical data has finished loading.
4Add SQL models from GitHub or dbt
Publish transformation models from a repository or connect an existing dbt or Cube project, then attach tests and check lineage.
5Orchestrate and activate the results
Chain syncs, builds and tests with dependencies and retries, then sync results back to CRMs, ad platforms or BI tools through reverse ETL.
6Connect an agent through the MCP server
Register the Weld MCP endpoint with an API key in Claude or Cursor, then ask the agent to inspect runs, build models or trigger syncs.

Pros and cons

Pros

Combines ingestion, SQL transformation, orchestration and reverse ETL in a single platform, with each piece usable on its ownAI
Exposes 60+ tools through an MCP server, so agents in Claude, Cursor or ChatGPT can create syncs, query models and inspect failed runsAI
Strong connector coverage spanning SaaS apps, databases, files and event sources, with CDC streaming and schema drift handlingAI
Transformation layer works with projects teams already own: dbt, Cube, SQLMesh and GitHub based modelsAI
Rated 4.8 on G2 across 109 reviews, with named customer stories from several European companiesAI

Cons

Querying is blocked until an initial sync finishes processing full history, which can delay time to first insight on large sourcesAI
No pricing information is published on the homepage, so cost has to come from a sales conversationAI
The agent and API story assumes comfort with MCP configuration, API keys and warehouse schemas, so non-technical users will struggleAI
Sources, destinations and connectors are visually listed but details such as per-connector limits or custom API support are not documented on the homepageAI
Scope is broad, which means teams wanting only a single lightweight sync may end up evaluating more platform than they needAI

How much does Weld cost?

Free trial

The homepage offers a Start for free link that opens workspace signup, but the duration or scope of any free tier is not stated.

Learn more

Integrations

Lists 300+ connectors plus warehouse destinations including BigQuery, Snowflake, Databricks, Redshift, ClickHouse and Postgres, BI tools such as Looker and Tableau, and AI clients including Claude, Cursor and ChatGPT.

Features

Covers ELT syncs with incremental, full and CDC replication from 300+ sources, SQL transformations via GitHub, dbt, Cube or SQLMesh, orchestration with dependencies and retries, reverse ETL to CRM and ad platforms, plus MCP server and REST API access for agents and applications.

Frequently asked questions about Weld

Does Weld offer a free trial?
The homepage offers a Start for free link that opens workspace signup, but the duration or scope of any free tier is not stated.
Does Weld have an API?
Yes, Weld offers an API.
How do you use Weld?
The walkthrough on this page covers 6 steps: 1. Create a workspace account 2. Connect a source and a warehouse destination 3. Run the initial sync 4. Add SQL models from GitHub or dbt 5. Orchestrate and activate the results 6. Connect an agent through the MCP server.
What platforms does Weld support?
Weld is available on Web App.
What does Weld integrate with?
Lists 300+ connectors plus warehouse destinations including BigQuery, Snowflake, Databricks, Redshift, ClickHouse and Postgres, BI tools such as Looker and Tableau, and AI clients including Claude, Cursor and ChatGPT.
What are the limitations of Weld?
Querying is blocked until an initial sync finishes processing full history, which can delay time to first insight on large sources. No pricing information is published on the homepage, so cost has to come from a sales conversation. The agent and API story assumes comfort with MCP configuration, API keys and warehouse schemas, so non-technical users will struggle.