Elysia

What it does

Elysia takes a natural-language question and, through a decision agent, chooses which tools to call in a decision tree. It can use custom Python tools you define with a decorator, or pre-built tools that query and aggregate data in a Weaviate cluster. Outputs include a response and retrieved objects from your collections.

People typically install it with pip as elysia-ai, then either run the local web app with elysia start and configure API keys, Weaviate cluster details and models in the settings page, or import the Tree class directly in Python. Before searching your own data, collections must be preprocessed, which the app does through an Analyze button. Its distinct approach is a predefined web of nodes orchestrated by a decision agent rather than exposing every tool at runtime.

Elysia suits developers and data teams who want to build agentic search over Weaviate data and can run Python 3.12 with model and vectoriser API keys or a local Ollama setup. Access is open source and free to use, but connected LLM and cluster usage depends on third-party providers. Key limitations: it is in beta, long context can make smaller local models time out or fail, and the project may not receive production-level maintenance.

What it can do

• Decision-agent routing across a predefined tree of tools
• Pre-built query and aggregate tools for Weaviate collections
• Custom Python tools registered with a tool decorator
• Preprocessing of Weaviate collections for agentic search
• Web app with settings for API keys, cluster details and models
• Local operation with local Weaviate and Ollama models

Works with

• Weaviate (cloud cluster or local instance)
• OpenRouter
• OpenAI
• Ollama

Example workflows

Query a Weaviate collection in Python

Illustrative example. Input: a question and a collection name such as Ecommerce. Steps: create a Tree, call it with the question and collection_names. Output: a response and matching objects from the collection.

Add a custom tool

Illustrative example. Input: a Python function with typed parameters. Steps: decorate it with @tool(tree=tree) and call the tree with a related question. Output: the decision agent can invoke the custom function and return its result.

Before you choose

• Elysia is beta and may not receive the maintenance level of a production service.
• Smaller local models can struggle with long context and structured outputs.
• Model, database and hosting costs depend on the providers or infrastructure you choose.

Listed under

• Data
• Agentic AI
• Chat interface
• Data exploration
• Developers
• Developer tools
• Ecommerce data

Get started

Run pip install elysia-ai, then elysia start or import the Tree class.

1. Install Python 3.12 and run pip install elysia-ai, optionally inside a virtual environment.
2. Set WCD_URL, WCD_API_KEY and WEAVIATE_IS_LOCAL in a .env file or the app settings, plus model keys such as OPENROUTER_API_KEY.
3. Run elysia start, open localhost:8000, and click Analyze in the Data tab after connecting a cluster to preprocess collections.

Elysia is free open-source software. Connected models, database services and hosting have their own costs and current allowances. Local Weaviate and Ollama setups are supported; no obsolete cloud-trial duration is assumed.

Elysia is free to use. See their pricing page ↗

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• GitHub - weaviate/elysia: Python package and backend for the Elysia platform app. · GitHub ↗github.com
• Elysia API ↗weaviate.github.io

Company website

weaviate.github.io: product, pricing and policy pages

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.