WeaveMindAI

weft

AIRust LLMProgramming LanguageWorkflow AutomationSvelte

+340

// summary

Weft is a programming language designed to integrate LLMs, human interactions, and infrastructure into a unified, visual workflow. It features durable execution to ensure programs survive crashes and supports complex logic through a typed, modular node system. Developers can build and manage sophisticated agentic systems by wiring together native nodes without the need for manual plumbing.

// technical analysis

Weft is a programming language specifically designed for AI-driven systems, aiming to replace complex library plumbing with a unified, graph-based architecture. By treating LLMs, human interactions, and infrastructure as first-class primitives, it allows developers to wire complex workflows together while the compiler enforces strict type safety and architectural integrity. The project leverages durable execution via Restate to ensure state persistence across crashes, prioritizing reliability in long-running processes like human-in-the-loop workflows.

// key highlights

Provides first-class support for human-in-the-loop workflows, allowing programs to pause and resume seamlessly without manual state management.

Features a recursively foldable design where complex systems can be collapsed into simple, interface-defined blocks for better readability.

Ensures end-to-end type safety with a compiler that validates connections and architecture before the code is executed.

Utilizes durable execution to guarantee that programs survive crashes and restarts, maintaining state across long-duration tasks.

Offers dual-view rendering, allowing developers to switch between dense code and visual graphs while keeping both views perfectly synchronized.

Includes a diverse catalog of built-in nodes for LLMs, communication platforms, databases, and web services to minimize external dependencies.

// use cases

Orchestrating AI agents and LLM workflows

Managing human-in-the-loop processes with persistent state

Building infrastructure-aware applications with native database and API integration

// getting started

To begin, clone the repository and configure your environment by copying the .env file and adding necessary API keys. Use the provided dev.sh script to launch the backend server and dashboard, then navigate to http://localhost:5173 to start building your first project. Consult the official documentation for a guided walkthrough of the language and dashboard features.