Skip to main content

Introduction

Qaynaq is a modern, open-source data pipeline tool designed to be a powerful and efficient alternative to tools like Airbyte and Fivetran. It empowers data analysts and scientists to easily build and manage data flows with a visual pipeline builder UI.

At its core, Qaynaq uses Bento as the flow processing engine. Bento provides battle-tested connectors, at-least-once delivery guarantees, and a rich set of processors — all driven by Qaynaq's coordinator-worker architecture and managed through the UI without writing any configuration by hand.

Key Features​

  • Visual Flow Builder — Intuitive UI to visually create and manage data pipelines with support for conditional routing and fan-out patterns.
  • Powerful In-Pipeline Transformations — Bloblang DSL for efficient data transformation and enrichment within the pipeline, replacing the need for separate tools like dbt.
  • MCP Integration — Expose flows as tools for AI assistants via the Model Context Protocol. Connect with Claude Desktop, Claude Code, Cursor, and other MCP-compatible clients.
  • Flexible Subprocess Processor — Integrate processors written in any programming language. Communication via stdin/stdout ensures language-agnostic compatibility.
  • Native HTTP Input — Accept data over HTTP, ideal for webhooks and flowing data sources.
  • Horizontally Scalable Worker Pool — Scale your data processing with a horizontally scalable worker pool architecture.
  • Delivery Guarantees — Reliable data delivery with buffering, caching, and robust error handling.

Why Qaynaq?​

  • Completely free — Apache 2.0 license with no usage limits or paid tiers.
  • Zero operational overhead — Runs as a single lightweight binary with no Docker, JVM, or external dependencies required.
  • Native transformations — The built-in Bloblang DSL handles mapping, filtering, and conditional logic, eliminating the need for separate tools like dbt.
  • MCP Server built-in — Turn any flow into an AI tool with the built-in Model Context Protocol server. No external infrastructure needed.
  • Language-agnostic processors — Write custom processors in any language through simple stdin/stdout communication.
  • Built-in observability — Metrics, tracing, and logs out of the box.
  • High performance — Written in Go with low memory and CPU footprint. Supports both real-time and batch workloads with fine-grained parallel execution control.

Who is Qaynaq for?​

  • Data Engineers who want a lightweight, self-hosted ETL tool without Docker overhead.
  • Backend Developers who need to move data between systems with transformations.
  • AI Developers building agentic systems that need custom tools for AI assistants via MCP.
  • Small Teams that want powerful data pipelines without enterprise complexity or cost.

Next Steps​