01 / PLOS ONE · 2008
Tool-using dolphins
Why Do Dolphins Carry Sponges?
Open document · PDF ↗ETHAN · Document intelligence
From your documents to answers with sources.
Explore a public adaptation of the ETHAN document experience. Fifteen publications about cetaceans to ask questions, check sources and keep learning.
This demonstration uses public documents and processes questions through Google Gemini. A private installation with local models is a different configuration.
CORBAT / ETHANES · EN
Interactive demo
What would you like to understand?
Whales and dolphins. Ask, check and keep exploring.
Library
Fifteen publications. One conversation.
02 / PLOS ONE · 2010
Whales and ocean nutrients
The Whale Pump: Marine Mammals Enhance Primary Productivity in a Coastal Basin
Open document · PDF ↗03 / Nature Communications · 2024
Sperm whale communication
Contextual and combinatorial structure in sperm whale vocalisations
Open document · PDF ↗Your knowledge, in your infrastructure
ETHAN supports document storage and indexing, generation and embeddings on a computer or infrastructure you control. With local models, document processing can run without sending documents or queries to external AI APIs.
Configuration must also cover tools, connections, logs and backups. We evaluate hardware, answer quality and execution times for each installation.
Discuss a private installationHow it is built
Ask, check and keep exploring.
Organize your questions into projects, return to your conversations and check each answer in its document. Open several sources, read the PDF at the cited page and consult an auxiliary translation when needed.
This demo uses fifteen scientific publications about cetaceans. History is saved only in this browser and you can export or delete it. Conversations expire after 30 days of inactivity. Questions are processed on our server and Google Gemini, with usage limits.
The platform and this demonstration
ETHAN uses Python, FastAPI, LangGraph and hybrid search with PostgreSQL/pgvector, with a React interface. This public adaptation uses React/TypeScript, Node.js, its own document retrieval and Gemini through Coco Runtime.
Knowledge and tools in your infrastructure
Integration conceptKnowledge and action in your own environment
We propose combining ETHAN’s visual interface and document knowledge with the tools of a locally installed Coco agent. This connection is not implemented yet.
ETHAN
Interface, library and authorized sources
Coco
Files and tools for authorized tasks
Local model
Generation and embeddings in your environment
- Question in ETHAN
- Source retrieval
- Task proposal
- Review or authorization
- Execution with Coco
- Result in the interface
For example: consult an internal procedure, identify the applicable steps and prepare a document or file changes from one interface.
With documents, index, embeddings and model in your infrastructure, information can be processed without external AI APIs. Tools, MCP, web search, logs and backups must also be configured for that goal.
How they would connect
A local adapter would exchange bounded requests and results while preserving ETHAN’s document permissions. Coco would not receive unrestricted database access or access to every document.