MindsDB
Unified open-source workspace to delegate projects and run AI agents.
Delegate entire projects to swappable open-source models and agents with secure data vaults.
MindsHub Cowork is an open-source platform superproject combining a desktop and web app, agent backends, and a data engine. It suits data scientists and professionals looking to automate multi-step workflows using connected data sources.
What it is
It connects secure data vaults like BigQuery, Postgres, and Notion to a model router supporting frontier and open-source models. You can run interchangeable open agents like Anton and Hermes, turning their output into published web artifacts.
Who it's for
It suits data workers and strategists wanting to orchestrate open models on their own data without vendor lock-in. It does not suit developers looking for low-level machine learning training frameworks or basic chat interfaces.
Key features
Secure Data Vault
Links systems like BigQuery and Postgres while keeping raw credentials hidden from agents.
Model Router
Switch between frontier and open-source models without configuring individual provider keys.
Open-Source Agents
Run interchangeable agent harnesses including Anton and Hermes directly from a dropdown.
Artifact Publishing
Turn raw agent outputs into live documents, dashboards, and applications with shareable URLs.
Up and running in 3 steps
Our honest verdict
What's great
- Supports swapping between proprietary and open-source models without changing built workflows
- Isolates external database and service credentials securely from the underlying agent runtimes
- Includes a platform superproject setup with Makefile commands for local and web development
- Can be deployed flexibly across cloud, VPC, on-prem, and air-gapped infrastructure
Where it falls short
- Requires managing submodules, Python environments, and local runtime dependencies when building from source
- Running local agents and database connectors demands dedicated system resources and technical configuration
- Advanced frontier models and private artifacts require a paid Pro subscription tier
Pricing & license
Free to start under the MIT license, with a Pro tier available that adds access to all frontier models and private artifacts.
Open remote AI-training roles
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Can I use my own API keys for models?
Yes, you can configure provider keys through the local runtime stored in your user directory.
How do I update submodule branches?
You can configure branches in a dev.env file and run make use to check them out across submodules.
What does the flush command do?
The make flush command removes local runtimes and app state databases for a completely fresh start.
