Hey Product Hunt, Excited to introduce you to Flow - a lightweight task engine for building AI agents. This launch is part of Mega Launch Week’s “Battle Royale" -> https://launchweek.dev/lw/MEGA. TLDR: Flow is designed to fix LangGraph's shortcomings. I think graph is a wrong abstraction for building AI agents. Just look at how incredibly hard it is to make routing using LangGraph - conditional edges are a mess. I built Laminar Flow to solve a common frustration with traditional workflow engines - the rigid need to predefine all node connections. Instead of static DAGs, Flow uses a dynamic task queue system that lets workflows evolve at runtime. Flow is built on 3 core principles: * Concurrent Execution - Tasks run in parallel automatically * Dynamic Scheduling - Tasks can schedule new tasks at runtime * Smart Dependencies - Tasks can await results from previous operations All tasks share a thread-safe context for state management. This task-based architecture makes complex workflows surprisingly simple: * Parallel task execution without explicit threading code * Self-modifying dynamic workflows and cycles * Conditional branching and control flow * Streaming of tasks execution * State management, load previous state and save current state * Start execution from a specific task * Dynamically push next tasks with specific inputs * Map Reduce, running the same task in parallel on multiple inputs and collecting results Flow is perfect for AI agents that need to make runtime decisions about their next actions. It is lightweight, elegantly written and has zero dependencies for the engine. The project also comes with built-in OpenTelemetry instrumentation for debugging and state reconstruction. Give it a try here -> https://github.com/lmnr-ai/flow. Just do pip install lmnr-flow. (or uv add lmnr-flow). More code examples are in the README. Flow is developed by the team behind Laminar (https://www.lmnr.ai)

Robert Kim

Laminar

Maker

📌

6mo ago

Luffy Lu

Congrats on launch! I love the departure from traditional static DAGs—dynamic task queues make so much sense for the unpredictability of AI-driven systems. With concurrent task execution and dynamic scheduling, how does Flow perform under heavy task loads or with highly parallel workflows? Have you benchmarked it against similar frameworks like LangGraph or Prefect?

Flow by Laminar [LW24]

open-source dynamic task engine for building AI agents

Do you use Flow by Laminar [LW24]?

Review Flow by Laminar [LW24]?