Lingo: The Architect in the Machine

Lingo is a minimal, async-native, and unopinionated toolkit for building modern LLM applications. It moves beyond generic "agents" and "chains" to focus on Conversational Modeling—the discipline of defining exactly how a system perceives, processes, and advances a dialogue state.

🧠 The Philosophy

Lingo is built on the principle that The Python stack is your state machine.

Most frameworks force you to define complex graphs, manage session IDs manually, or use "black-box" agents that are hard to steer. Lingo takes the opposite approach: it uses standard Python async/await to pause and resume execution, making the conversation history a first-class citizen of your code.

We call this The Architect in the Machine: a design pattern where the LLM is not just a chatbot, but an active component in a structured, typed architecture.

⚡ Core Values

• 💾 Stateful by Default: Use await engine.ask() to pause execution. Lingo automatically suspends the stack and preserves variables in memory across turns.
• 🛡️ Type-Safe: Built on Pydantic. All inputs, outputs, and tool calls are validated against your schemas.
• 🌊 Multi-Paradigm: Mix rigid business rules (States) with flexible reasoning (Skills) and event-driven guardrails (Reflexive Patterns).
• 🏗️ Unopinionated: No forced "agent" abstractions. Use as much or as little of the framework as you need.

🚀 Quick Start

To get a feel for Lingo, check out the User Guide for step-by-step recipes, or the Deployment Guide to set up your environment.

pip install lingo-ai

The "Hello World" (Stateful Wizard)

import asyncio
from lingo import Lingo

app = Lingo("Wizard")

@app.skill
async def onboarding(ctx, eng):
    await eng.reply(ctx, "Welcome!")
    # PAUSE execution and wait for user input
    name = await eng.ask(ctx, "What is your name?")
    # Resume and use context from previous turns
    await eng.reply(ctx, f"Hi {name}, nice to meet you!")

if __name__ == "__main__":
    from lingo.cli import loop
    loop(app)