Databench
Product · Self-hosted / on-prem software · Data & analytics by Alkera
The open-source, self-hostable data workspace for humans and agents.
Databench is an open-source, self-hostable workspace for live collaboration between people and AI agents on data tasks. It features reactive Python notebooks and SQLAlchemy connectivity, allowing teams to run a governed data environment on their own infrastructure.
- Languages
- English
Category: Open Source
- Deployment
- Self-hostable, on-premises, host machine, or over SSH
- Platforms
- Web-based
- Integrations
- Anthropic, OpenAI, SQLAlchemy-compatible databases
- License
- Apache-2.0
- Notebook File Extension
- .alknb.py
- Notebook Compatibility
- marimo
- Open-source Apache-2.0 license
- Self-hostable, web-based data workspace
- Reactive SQL and Python notebooks compatible with marimo
- Connects to any SQLAlchemy-compatible database
- Supports bring-your-own-keys for Anthropic and OpenAI
Overview
Databench by Alkera is an open-source, self-hostable data workspace designed for collaborative data engineering and analysis. It provides a web-based environment where human teams and AI agents work together in real-time across shared chats, files, and notebooks. The platform features reactive SQL and Python notebooks built on an open format compatible with marimo, ensuring that notebooks remain readable as plain Python files. Databench allows organizations to maintain control over their data by running on their own infrastructure and using their own API keys for models from providers like Anthropic and OpenAI. It supports connections to any SQLAlchemy-compatible database and provides administrative controls including nested teams and role-based access.
Collaboration and Workspace
- Shared Agent Chats: Teammates and an agent share each chat live for collaborative data work.
- Shared File System: The workspace includes shared files and folders for team access.
- Recorded Authorization: Every authorization decision and action is recorded for audit purposes.
Notebooks and Analysis
- Reactive Python Notebooks: Notebooks are reactive and written in plain Python, compatible with the marimo format.
- SQL and Python Integration: Notebooks support a mix of SQL and Python cells for data analysis.
- Tables and Charts Renderer: The platform includes a built-in renderer for displaying tables and charts within notebooks.
- Open Notebook Format: Notebooks are saved as.alknb.py files that diff cleanly in Git and run as scripts.
Infrastructure and Connectivity
- SQLAlchemy Connections: Connect to any database that is compatible with SQLAlchemy.
- Flexible Compute Options: Run chats and notebooks on the host machine or add external machines over SSH.
- Bring Your Own Model Keys: A model gateway calls Anthropic or OpenAI using your own API keys.
- Sandboxed Agent Execution: The agent runs in its own sandbox on a machine chosen by the user.
Use cases
- Data teams use Databench to collaborate live with AI agents on SQL and Python notebooks for rapid analysis.
- Organizations deploy Databench on-premises to keep sensitive data and AI interactions within their own network.
- Analysts use the reactive notebook environment to create interactive charts and tables from SQLAlchemy-compatible databases.
Specifications
- Deployment
- Self-hostable, on-premises, host machine, or over SSH
- Platforms
- Web-based
- Integrations
- Anthropic, OpenAI, SQLAlchemy-compatible databases
- Sign-in and administration
- Password sign-in, nested teams, invitations, roles, and recorded authorization decisions
- License
- Apache-2.0
- Notebook File Extension
- .alknb.py
- Notebook Compatibility
- marimo
- Database Compatibility
- SQLAlchemy-compatible
- AI Model Providers
- Anthropic, OpenAI
- Deployment Options
- Self-hosted, SSH-connected machines
- Collaboration Features
- Shared chats, shared files, live editing
How it works
- Download and Deploy: Users can access the source code on GitHub to deploy the platform on their own infrastructure.
- Configure Connections: Connect to existing data sources using SQLAlchemy-compatible drivers and configure AI model keys.
- Collaborate and Analyze: Invite team members, set up roles, and start collaborating in shared chats and notebooks.
FAQ
What is the license for Databench?
Databench is licensed under the Apache-2.0 open-source license.
What notebook format does Databench use?
It uses an open-source format based on marimo, where notebooks are saved as plain Python (.alknb.py) files.
Which databases can I connect to?
You can connect to any database that is compatible with SQLAlchemy.
Can I use my own AI model keys?
Databench includes a model gateway that allows you to use your own keys for Anthropic or OpenAI.
Added by AntSpace from public sources (alkera.ai).