Elysia: Decision Tree Agentic Framework

2026-05-14 · source · source ↗ · raw capture · wiki page

Elysia is an agentic platform from Weaviate that organizes tool usage through a decision tree architecture. A "decision agent" selects tools dynamically based on environmental context. It supports custom tools as well as pre-built tools for interacting with Weaviate clusters.

Installation

Requires Python 3.12. Install via pip install elysia-ai. Also available from GitHub for development.

Quickstart

App: elysia start then open localhost:8000. Configure API keys, Weaviate cluster details, and models in the settings page.

Python:

from elysia import tool, Tree

tree = Tree()

@tool(tree=tree)
async def add(x: int, y: int) -> int:
    return x + y

tree("What is the sum of 9009 and 6006?")

With Weaviate:

import elysia
tree = elysia.Tree()
response, objects = tree(
    "What are the 10 most expensive items in the Ecommerce collection?",
    collection_names = ["Ecommerce"]
)

Architecture

Backend: FastAPI. Core logic: pure Python with DSPy handling LLM interactions. Each tree node is orchestrated by a decision agent with global awareness of environment, available actions, and past/future steps.

Supports OpenAI, OpenRouter, and local models via Ollama.

Stats