Most ETL tools require either writing code (Airflow, Dagster) or paying enterprise pricing (Fivetran, Hightouch). QueryFlow's Visual ETL builder lets you build production pipelines by dragging boxes onto a canvas. AI field mapping handles the tedious part. Native Mac, $299/year.
Quick answer: QueryFlow's Visual ETL builder lets you build ETL pipelines without code. Drag a source onto the canvas, drag a destination, click the connecting line to open the Field Mapper. AI Map auto-detects field matches using 25+ synonym patterns. Add transform chips for case changes, date formats, custom Python. Schedule the pipeline to run on cron. $299.99/year on macOS.
Most no-code ETL tools fall into two categories. Cloud SaaS (Fivetran, Hightouch, Stitch, Airbyte Cloud) — managed, expensive, vendor lock-in. Browser-based low-code (n8n, Make, Zapier) — flexible but slow for large data volumes. Neither category serves the developer who wants no-code convenience with desktop-app performance and reasonable pricing. QueryFlow fills that gap.
Open a new pipeline. Drag a source card from the sidebar (Snowflake, Postgres, MySQL, Redshift, Google Sheets, Salesforce, or CSV) onto the canvas. Configure the source: pick a table or write a query. Drag a destination card. Configure the destination. Click the connecting line between source and destination to open the Field Mapper. Map fields visually with bezier curves. Hit AI Map for auto-matching. Save. Run. Optionally schedule.
The Field Mapper has an AI Map button that auto-matches source columns to destination fields using a synonym dictionary with 25+ common patterns. Examples: email/email_address/EmailAddress → Email. fname/first_name/FirstName → FirstName. company/account/AccountName → AccountName. customer_id/customerId/external_id → ExternalId. The patterns cover the vast majority of real-world ETL field mapping work. For non-obvious cases, Claude AI's 'Suggest Mappings' button uses your actual schemas to make smart suggestions.
Between source and destination fields, you can add transform chips: UPPERCASE, lowercase, TRIM, date format conversion (multiple formats supported), CONCAT (combine multiple source fields), SPLIT (extract part of a source field), DEFAULT (replace nulls with a value), and Custom Python for arbitrary transformations. Chips stack — you can UPPERCASE then TRIM a field. Each chip has a small UI for configuration.
Some transformations are genuinely complex — deduplication logic, multi-table joins, conditional logic based on source data. For these, Flow Books handle the work: SQL cells to query source data, Python cells with pandas to transform, output goes to the same destinations the Visual ETL supports. Flow Books and Visual ETL pipelines coexist — use whichever is right for the specific workflow.
Building a pipeline in Visual ETL: 5-10 minutes for a typical Salesforce import or warehouse sync. Building the same pipeline in code (Python script with libraries, SQL deployment, scheduler setup): 1-3 hours. The time savings compound when you have many small pipelines (which most teams do).
Built pipelines can be scheduled with cron, interval, daily, weekly, or monthly triggers. The scheduler runs locally on your Mac via macOS SMAppService. Failed runs surface in the Observatory dashboard with full error context and one-click retry. The pipeline you build today runs every morning for years without intervention.
Visual ETL pipelines are production-grade. Same Bulk API v2 for Salesforce as code-based tools. Same COPY INTO for Snowflake bulk loads. Same wire-protocol connections for Postgres and MySQL. The 'no-code' is in the build experience, not the underlying execution — the runtime is just as capable as code-based alternatives.
Yes for the vast majority of workloads. The thing 'no-code' actually removes is the build/maintain overhead — the actual data movement uses the same underlying APIs and protocols as code-based tools. QueryFlow users run pipelines moving millions of rows daily through Visual ETL pipelines.
Use a Custom Python chip, or switch the transformation into a Flow Book. Flow Books give you full pandas and the Python ecosystem for arbitrarily complex transformations, then write to the same destinations.
Pipeline definitions are stored in QueryFlow's local SwiftData store. Cross-machine version control isn't a v1.5 feature, but it's planned via iCloud sync (on the public roadmap with active voting). For now, you can export pipeline definitions manually for backup.
AI Map typically gets 80-90% of mappings correct automatically. The Field Mapper UI makes it easy to override any auto-mapped pair with a click. For complex schemas with non-standard naming, Claude AI's 'Suggest Mappings' adds context-aware reasoning beyond simple synonym matching.
v1.5 has one destination per pipeline. To send the same source data to multiple destinations, you can either run two pipelines on the same schedule, or use a Flow Book that explicitly writes to multiple destinations programmatically.
14-day free trial. Build your first visual pipeline in 10 minutes and see how much faster ETL development gets.